MAINTENANCE and DYNAMICS of RAINFOREST EDGES Yong Tang

Maintenance and Dynamics of Rainforest Edges

Author Tang, Yong

Published 2008

Thesis Type Thesis (PhD Doctorate)

School School of Environment

DOI https://doi.org/10.25904/1912/3066

Downloaded from http://hdl.handle.net/10072/367442

Griffith Research Online https://research-repository.griffith.edu.au

MAINTENANCE AND DYNAMICS OF RAINFOREST EDGES

Yong Tang

2007

MAINTENANCE AND DYNAMICS OF RAINFOREST EDGES

Yong Tang B.Sc, M.Sc. (CAS)

Griffith School of Environment Faculty of Environmental Sciences and Engineering Griffith University

Supervisors Prof. Roger L. Kitching Dr. Jacinta Zalucki

Submitted in fulfillment of the requirements of the degree of Doctor of Philosophy October 2007

SYNOPSIS

Deforestation and fragmentation of rainforest has become one of the major threats to global biodiversity and the massive loss of rainforest during the past decades has pushed the global biota to the edge of the global species extinction crisis. Despite the increasing public awareness and tremendous efforts made internationally to save the remaining rainforest, the deforestation rate continues to accelerate in many rainforest areas. This trend is due mainly to increasing human population and local or regional economical or political crises creating increased needs and demands on land use and rainforest products. In addition to the loss of large areas of wildlife habitat, a direct consequence of rainforest fragmentation is the increase in the extent of edges, through which “hostile” edge effects can have a profound impact on the dynamics of remaining rainforests. There is an urgent need to understand the underlying mechanisms that drive the dynamics of the rainforest edges and more important, the subsequent long-term impact on the local and regional rainforest.

The main objective of the study described in this thesis has been to compare the patterns with which rainforest plants respond to the edge environment at different types of edges involving rainforests. The study was conducted within a fragmented subtropical rainforest complex in Lamington National Park, Southeast Queensland. Rainforest trees, lianas, seedling banks and soil seed banks were investigated at eucalypt forest/ rainforest, pasture/ rainforest and roadside rainforest edges. For each edge type, nine 100 m transects were established from the edge to rainforest interior and transects were extended 50 m into eucalypt forest and pasture for additional sampling of surrounding matrices. Vegetation surveys were conducted along the edge transects for the study of trees, lianas and seedlings. Soil seed banks were investigated by germination experiments conducted in a shade house, using soil samples collected along the edge transect. The results from the edge studies were compared with corresponding studies in a 1 ha rainforest reference plot located in a relatively undisturbed area within the rainforest interior.

Studies focussed on the two factors to which rainforest plants near the edges respond. 1) Edge to rainforest interior gradients, driven by changes in microclimatic factors

and associated modified species composition, structure and ecological processes within rainforest near the edges. 2) The different disturbance regimes associated with the matrices that surround the edges. Both univariate and multivariate statistical approaches have been used to describe the underlying patterns and to identify plant species that respond to these patterns.

The results demonstrated that the species composition and structure of rainforest has been largely modified near the edges in response to edge types and associated disturbances. These changes include a dramatic increase in the densities of small trees and lianas, a shift of shade-tolerant species to shade-intolerant species and a significant increase in the presence of fleshy-fruited plants. In addition, canopy lianas also showed a significant shift from twiners to tendril climbers. The distance from edge to rainforest interior over which the changes in forest composition and structure can be recognized, varied among edges, from a few metres at roadside rainforest edges to more than 80 metres at pasture/ rainforest edges. The results also suggest that the eucalypt forest/ rainforest edges have apparently developed a unique edge plant assemblage in response to long-term fragmentation and repeated burning.

The changes in the forest canopy have also lead to corresponding changes in seedling banks that depend primarily on the species and microclimatic conditions maintained within the rainforest canopies. A significant decline in the density of rainforest tree seedlings was found at rainforest edges, compared with the rainforest reference plot. This decline followed a gradient from rainforest interior to the edges at eucalypt/ rainforest and pasture/ rainforest edges. In contrast, a dramatic increase in the density of liana seedlings was found at all three types of edges. The densities of liana and tree seedlings were found to be positively correlated within the rainforest reference plot, but this relationship became negative at rainforest edges. This implies that the edge effects may potentially favour the regeneration of lianas.

A possible seed dispersal limitation has been suggested by the results from soil seed bank studies. This dispersal limitation may have contributed to a significant decrease of seed storages of rainforest species in the soil seed banks from the rainforest interior towards the edges and continued through to the surrounding matrices. Not a single ii

native tree seed was germinated from soil samples taken within the pastures. Furthermore, large amount of seeds originating from the surrounding matrices have been brought into the rainforest. The distribution patterns of those non-rainforest seeds, decreasing linearly from edge to rainforest interior, has demonstrated the function of edges as an effective barrier preventing the penetration of species and material from the surrounding matrices. The analysis of the relationship between the soil seed bank and above-ground vegetation has emphasized the importance of ‘long- lived’ pioneer species in the dynamics of the subtropical rainforest and their potential for rainforest restoration.

This study also suggests that a prolonged drought may have had a profound impact on the dynamics of the rainforest in the study area. These impacts include the drought- induced mass-fruiting of some species, such as Caldcluvia paniculosa and Derris involuta, and changes in the microclimatic conditions for seedling establishment, causing significant changes in the soil seed bank and seedling bank. A potential interaction of drought and edge effects may lead to a significant increase in liana seedlings at the expense of tree seedlings.

This study also has discussed Australian rainforest fragmentation in a geological context with consideration of human induced disturbance prior to and after European settlement. This suggests that the Australian rainforest may be more resistant to fragmentation due to long term “adaptation” through repeated fragmentation across geological time and the absence of large scale agriculture prior to the European settlement. On the other hand, the high endemism and the long term isolation in a sea of highly flammable sclerophyllous vegetation suggest that the Australian rainforest remnants may be more vulnerable to projected climate change. The short-term intensive rainforest destruction after European settlement has contributed substantially to this vulnerability.

Findings from this study have important implications for long-term trends in rainforest dynamics. A lack of large areas of secondary rainforest may become an obstacle to natural regeneration in deforested areas and compromise the conservation value of the remaining rainforest remnants. Rainforest edges are one of only a few places where large numbers of secondary rainforest species, especially those of

iii

fleshy-fruited trees and lianas, are well preserved. The edges may become important seed sources promoting natural regeneration in deforested areas. The existence of dispersal limitation near the edges, however, may limit the functioning of this natural service. Long-term monitoring of the dynamics of rainforest edges is needed which can be achieved by using rapid assessment of soil seed banks near the edges.

STAEMENT OF ORIGINALITY

I certify that this thesis is my original work and has not previously been submitted for a degree or diploma in any university. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made in the thesis itself.

Yong Tang

TABLE OF CONTENTS

SYNOPSISTU UT ...... i

STAEMENTTU OF ORIGINALITYUT ...... v

TABLETU OF CONTENTSUT ...... vii

LISTTU OF FIGURESUT ...... x

LISTTU OF APPENDICESUT ...... xvii

ACKNOWLEDGEMENTSTU UT ...... xviii

COLOURTU PLATESUT ...... xxi

1TU UT GENERALTU INTRODUCTIONUT ...... 1

1.1TU UT TheTU Global Rainforest CrisisUT ...... 1

1.2TU UT RainforestTU Fragmentation and Edge EffectsUT ...... 1

1.3TU UT TheTU Future of Rainforest RemnantsUT ...... 4

1.4TU UT RainforestTU Fragmentation: an Australian PerspectiveUT ...... 5

1.4.1TU UT RainforestTU fragmentation in a geological contextUT ...... 6

1.4.2TU UT AboriginesTU and rainforest fragmentationUT ...... 6

1.4.3TU UT EuropeanTU impacts on rainforestUT ...... 8

1.4.4TU UT AustralianTU rainforest in a changing climateUT ...... 8

1.5TU UT RainforestTU Edge Studies in AustraliaUT ...... 9

1.6TU UT ObjectivesTU of This ThesisUT ...... 10

1.7TU UT StructureTU of This ThesisUT ...... 11

2TU UT STUDYTU AREA AND METHODOLOGYUT ...... 13

2.1TU UT LocalityTU and GeographyUT ...... 13

2.2TU UT ClimateTU UT ...... 13

2.3TU UT VegetationTU UT ...... 16

2.4TU UT TheTU History of Lamington National ParkUT ...... 16

2.5TU UT StudyTU Area and SitesUT ...... 18

2.5.1TU UT ReferenceTU rainforestUT ...... 18

2.5.2TU UT Rainforest/TU eucalypt forest edge...... 19UT

2.5.3TU UT Rainforest/TU pasture edgeUT ...... 21

2.5.4TU UT RoadsideTU rainforest edgeUT ...... 22

2.6TU UT VegetationTU SurveyUT ...... 22

2.6.1TU UT EdgeTU vegetation surveyUT ...... 22

2.6.2TU UT TheTU reference rainforest surveyUT ...... 24

2.7TU UT SoilTU Seed Bank SamplingUT...... 26

2.8TU UT GerminationTU ExperimentUT ...... 26

2.9TU UT DataTU AnalysisUT ...... 27

3TU UT RAINFORESTTU FRAGMENTATION AND TREESUT ...... 29

3.1TU UT IntroductionTU UT ...... 29

3.2TU UT MethodsTU ...... 31UT

3.2.1TU UT TreeTU samplingUT ...... 31

3.2.2TU UT DataTU analysisUT ...... 31

3.3TU UT ResultsTU UT ...... 33

3.3.1TU UT RainforestTU in the one-hectare plotUT ...... 33

3.3.2TU UT ChangeTU of tree assemblages at rainforest edgesUT ...... 34

3.3.3TU UT DistributionTU of tree densityUT ...... 35

3.3.4TU UT DistributionTU of tree basal areaUT ...... 37

3.3.5TU UT SimilarityTU among different size classesUT ...... 39

3.3.6TU UT DistributionsTU of trees in different regeneration guildsUT ...... 40

vii

3.3.7TU UT CompositionTU of shade-intolerant speciesUT ...... 44

3.4TU UT DiscussionTU UT ...... 44

3.4.1TU UT MatrixTU regulated edge effectsUT ...... 46

3.4.2TU UT TheTU distance effectsUT ...... 47

3.4.3TU UT Shade-intolerantTU trees at rainforest edgesUT ...... 48

3.4.4TU UT TheTU future of rainforest edgesUT ...... 49

4TU UT LIANATU AND TREE ASSOCIATIONS IN A FRAGMENTED

SUBTROPICAL RAINFORESTUT ...... 51

4.1TU UT IntroductionTU UT ...... 51

4.2TU UT MethodsTU ...... 53UT

4.2.1TU UT LianaTU samplingUT ...... 53

4.2.2TU UT DataTU analysisUT ...... 53

4.3TU UT ResultTU UT ...... 57

4.3.1TU UT LianaTU compositionUT ...... 57

4.3.2TU UT ChangesTU in liana densityUT ...... 59

4.3.3TU UT ChangeTU in liana basal area...... 61UT

4.3.4TU UT ComparisonTU of liana compositionUT ...... 63

4.3.5TU UT ComparisonTU of liana climbing guildsUT ...... 65

4.3.6TU UT ComparisonTU of liana dispersal guildsUT ...... 66

4.3.7TU UT TreeTU and liana associationsUT ...... 68

4.4TU UT DiscussionTU UT ...... 72

4.4.1TU UT LianaTU distribution at rainforest edgesUT ...... 73

4.4.2TU UT MatrixTU regulated liana communities?UT...... 73

4.4.3TU UT LianaTU and tree associationUT ...... 74

4.4.4TU UT ChangeTU in liana climbing guildsUT ...... 77

4.4.5TU UT ChangeTU in liana dispersal guildsUT ...... 77

5TU UT THETU SEEDLING BANK AND RAINFOREST FRAGMENTATIONUT...... 79

5.1TU UT IntroductionTU UT ...... 79

5.1TU UT MethodTU UT ...... 82

5.1.1TU UT SeedlingTU bank sampling at the one-hectare reference plotUT...... 82

5.1.1TU UT SeedlingTU bank sampling at rainforest edgesUT ...... 82

5.1.2TU UT DataTU analysisUT ...... 82

5.2TU UT ResultTU UT ...... 84

5.2.1TU UT TheTU seedling bank within the one-hectare reference plotUT ...... 84

5.1.1TU UT TheTU seedling bank at rainforest edgesUT ...... 86

5.2TU UT DiscussionTU UT ...... 94

5.2.1TU UT Matrix-regulatedTU seedling banksUT ...... 94

5.2.2TU UT SeedlingTU banks and the forest canopyUT ...... 96

5.2.3TU UT ResponseTU of tree seedling to edge effectsUT ...... 97

5.2.4TU UT LianaTU and tree seedling associationsUT ...... 99

6TU UT THETU SPATIAL DISTRIBUTION OF THE SOIL SEED BANK AND

CANOPY TREES IN A SUBTROPICAL RAINFORESTUT ...... 103

6.1TU UT IntroductionTU UT ...... 103

6.2TU UT MethodsTU ...... 104UT

6.3TU UT ResultsTU UT ...... 105

6.3.1TU UT SummaryTU of the germination experimentUT ...... 105

6.3.2TU UT SoilTU seed bank and above-ground vegetationUT...... 107

6.3.3TU UT SpatialTU distribution of seeds in the soil seed bankUT ...... 109

6.3.4TU UT DistributionTU of seeds and canopy trees of Caldcluvia paniculosaUT .....112

6.4TU UT DiscussionTU UT ...... 114 viii

6.5TU UT ConservationTU applicationsUT ...... 118

7TU UT SOILTU SEED BANKS AT RAINFOREST EDGESUT ...... 119

7.1TU UT IntroductionTU UT ...... 119

7.2TU UT MethodsTU ...... 121UT

7.2.1TU UT DataTU collectionUT ...... 121

7.2.2TU UT DataTU analysisUT ...... 122

7.3TU UT ResultsTU UT ...... 123

7.3.1TU UT SummaryTU of germination experimentsUT ...... 123

7.3.2TU UT CompositionTU of soil seed banksUT ...... 125

7.3.3TU UT SpeciesTU estimationUT ...... 127

7.3.4TU UT SeedTU distribution along edgesUT ...... 128

7.3.5TU UT ResponseTU of individual species to distance from the edgeUT ...... 132

7.3.6TU UT ComparisonTU of soil seed bank composition among edgesUT ...... 133

7.3.7TU UT SoilTU seed bank and the rainforest canopyUT ...... 134

7.3.8TU UT SoilTU seed bank and seedling bankUT ...... 137

7.4TU UT DiscussionTU UT ...... 140

7.4.1TU UT Edge-regulatedTU soil seed bankUT ...... 141

7.4.2TU UT ResilienceTU of rainforest to disturbanceUT ...... 141

7.4.3TU UT DispersalTU limitation of rainforest seeds in fragmented landscapesUT ....146

7.4.4TU UT ConservationTU applicationsUT ...... 148

8TU UT GENERALTU DISCUSSIONUT ...... 149

8.1TU UT SummaryTU of FindingsUT ...... 149

8.1.1TU UT ChangesTU in the rainforest canopyUT ...... 149

8.1.2TU UT ChangesTU in the soil seed bank and seedling bankUT ...... 150

8.1.3TU UT EdgesTU under different disturbance regimesUT ...... 151

8.2TU UT TheTU Impact of the Prolonged DroughtUT ...... 153

8.3TU UT UsingTU Soil Seed Bank as Indicators of Edge EffectsUT ...... 154

8.4TU UT RegenerationTU Strategy of Rainforest PlantsUT ...... 156

8.5TU UT RainforestTU Edge: Future for Australian RainforestUT ...... 160

8.6TU UT FutureTU Research NeedsUT ...... 160

REFERENCESTU UT ...... 163

APPENDIXTU 1UT ...... 193

APPENDIXTU 2UT ...... 205

APPENDIXTU 3UT ...... 209

LIST OF FIGURES

FigureTU 2.1 The location of Lamington National Park and the study areaUT ...... 14

FigureTU 2.2 The monthly rainfall and average temperature of the study area.UT ...... 15

FigureTU 2.3 Long term patterns of the average rainfall of the study areaUT ...... 15

FigureTU 2.4 Layout of vegetation sampling plots along an edge transect.UT ...... 25

FigureTU 3.1 Species discovery curves based on first-order Jack-knife estimations of the

tree species richnessUT ...... 35

FigureTU 3.2 Changes in the densities of trees from edges to rainforest interiors, with

comparisons with the rainforest reference plotUT ...... 36

FigureTU 3.3 Distributions of the total basal area of trees from the edges to rainforest

interiors, with comparisons with the rainforest reference plotUT ...... 40

FigureTU 3.4 Distributions of the densities of trees in three regeneration guilds from edges to rainforest interiors, with comparisons with the rainforest reference

plotUT ...... 41

FigureTU 3.5 Distributions of the basal areas of trees in three regeneration guilds from edges to rainforest interiors, with comparisons with the rainforest reference

plotUT ...... 43

FigureTU 3.6 NMDS analysis based on the composition of shade-intolerant tree species at different edge types. Vectors show species significantly associated with the

ordination coordinates.UT ...... 45

FigureTU 4.1 Liana species accumulation curves based on first order jack-knife

estimations for the rainforest edges and the rainforest reference plot.UT ...... 59

FigureTU 4.2 Distributions of liana densities from edges to rainforest interiors, with

comparisons with the rainforest reference plotUT ...... 60

FigureTU 4.3 The distributions of liana basal area from edges to rainforest interior, with

comparisons with the rainforest reference plotUT ...... 62

FigureTU 4.4 MDS comparisons of liana composition among rainforest edges, with vectors showing increase in the abundance of liana species significantly

associated with the ordination axes.UT...... 64

FigureTU 4.5 Comparisons of the density and proportion of lianas in four climbing

guilds among rainforest edges and the rainforest reference plot.UT ...... 66

FigureTU 4.6 Comparisons of the density and proportion of lianas in different seed

dispersal guilds between rainforest edges and the rainforest reference plot.UT ...67 x

FigureTU 4.7 Observed numbers of lianas carried by rainforest trees within the rainforest reference plot and rainforest edges, fitted with expected Poisson distribution

valuesUT ...... 70

FigureTU 5.1 Number of seedlings in different size classes at three rainforest edges and

within the rainforest reference plot.UT ...... 87

FigureTU 5.2 Changes in the density of tree and liana seedlings from edges to rainforest

interiors, with comparisons with rainforest reference plot.UT ...... 89

FigureTU 5.3 MDS results for the composition of tree seedlings among rainforest edges,

with vectors showing species significantly associated with ordination axesUT ...90

FigureTU 6.1 Species accumulation curve based on the observed species richness in the soil seed bank of the 1 ha subtropical rainforest plot, fitted with a first-order

jackknife estimate.UT ...... 107

FigureTU 6.2 Spatial distributions of seeds in the soil seed bank within the 1 ha

subtropical rainforest plot.UT ...... 111

FigureTU 6.3 Spatial distribution of the canopy trees of Caldcluvia paniculosa within

the 1 ha subtropical rainforest plot.UT...... 112

FigureTU 6.4 Distribution patterns of seeds in relation to large sized canopy trees of

Caldcluvia paniculosa within the 1 ha subtropical rainforest plotUT ...... 113

FigureTU 6.5 Correlation of the seed density of Caldcluvia paniculosa in the soil seed

bank with the distances to the nearest conspecific large canopy tree UT ...... 114

FigureTU 7.1 Species accumulation curves based on first-order jackknife estimations on the seedlings germinated from the soil seed banks of rainforest edges and the

surrounding matrices.UT ...... 128

FigureTU 7.2 Distributions of seeds in the soil seed banks near three types of rainforest

edges.UT ...... 129

FigureTU 7.3 Distributions of exotic seeds in the soil seed banks near rainforest edges.UT ...... 130

FigureTU 7.4 Distributions of native seeds in the soil seed banks near rainforest edges.UT ...... 131

FigureTU 7.5 Distributions of tree seeds in the soil seed banks near rainforest edges.UT .131

FigureTU 7.6 MDS analysis of rainforest edges based on the overall seedlings germinated from the soil seed banks. The vectors indicate increases in the

density of species showing significant associations with the ordination axesUT ...... 133

FigureTU 7.7 MDS analysis of rainforest edges based on the tree seedlings germinated

from the soil seed banks. The vectors indicate an increase in density of species

that showed significant associations with the ordination axesUT...... 135

FigureTU 7.8 Proportions of trees in different size-classes of species shared by the soil

seed banks and the above-ground vegetation at the rainforest edges.UT ...... 138

xii

LIST OF TABLES

TableTU 2.1 Summary of sampling efforts for rainforest edges and the rainforest

reference plot UT ...... 20

TableTU 3.1 Density, species richness, and diversity of trees within the rainforest

reference plot and the rainforest edges.UT ...... 34

TableTU 3.2 ANOVA results of the comparisons of the densities of trees in different size classes at different positions from edges to the rainforest interiors and among

different types of rainforest edgesUT ...... 38

TableTU 3.3 Correlation coefficients between the densities and total basal area of trees

with the distances from edge to the rainforest interiorsUT ...... 39

TableTU 3.4 Sorensen Similarity Indices between tree assemblages in different size

classesUT ...... 39

TableTU 3.5 Results of two-way ANOVA for the densites of trees in different regeneration guilds at different distances from edges to rainforest interiors and

among different types of rainforest edges.UT ...... 42

TableTU 3.6 Spearman Rank Correlations between densities and total basal areas of trees in different regeneration guilds with the distances from edges to rainforest

interiorsUT ...... 44

TableTU 3.7 ANOSIM results for the comparisons of shade-intolerant species among

different types of rainforest edgesUT ...... 44

TableTU 4.1 Classifications of the climbing and seed dispersal guilds of lianasUT...... 55

TableTU 4.2 Density and total basal area of lianas and the infestation rate of trees at

rainforest edges and the rainforest reference plot.UT ...... 58

TableTU 4.3 Results of the two-way ANOVA for the densities and total basal area of lianas at different positions from edges to rainforest interiors and among three

types of rainforest edgesUT ...... 61

TableTU 4.4 Spearman Rank Correlations of density and total basal area of lianas with

the distances from the edges to the rainforest interiorsUT...... 63

TableTU 4.5 ANOSIM results for the compositions of lianas at different positions from

edges to rainforest interior and among the rainforest edgesUT ...... 63

TableTU 4.6 Results of the two-way ANOVA for the densities of lianas in different climbing guilds from edges to the rainforest interiors among different types of

rainforest edges.UT ...... 65

xiii

TableTU 4.7 Results of the two-way ANOVA for the density of lianas in different dispersal guilds at different positions from edges to the rainforest interiors and

among different types of rainforest edgesUT ...... 67

TableTU 4.8 The associations of dominant tree species with lianas (trees with> 30 stems

or carrying more than 20 lianas in the survey are listed).UT ...... 69

TableTU 4.9 Associations of dominant liana species with canopy trees.UT ...... 71

TableTU 4.10 Correlations between lianas and trees at rainforest edges and within the

rainforest reference plot.UT ...... 71

TableTU 4.11 Results of the Mantel tests for the comparisons of the similarity matrices

between assemblages of trees and lianas at rainforest edgesUT ...... 72

TableTU 5.1 Number of individuals and relative densities of the seedlings of the dominant species in the seedling bank and their relative densities in the tree

canopy in the rainforest reference plot...... 84UT

TableTU 5.2 Number of individuals and relative densities of lianas in the seedling banks

and their relative density in canopy lianas within the rainforest reference plot.UT ...... 85

TableTU 5.3 Correlations between tree and liana seedlings and canopy trees and lianas in

the 1 ha rainforest reference plotUT ...... 86

TableTU 5.4 Densities and species richness of tree and liana seedlings at rainforest

edges.UT ...... 88

TableTU 5.5 Results of the two-way ANOVA for the seedling densities at different

distances from edges to the rainforest interiors and among the rainforest edgesUT ...... 88

Table5.6TU ANOSIM results comparing the composition of tree seedling banks among

the rainforest edgesUT ...... 90

TableTU 5.7 Correlations between the densities of tree seedlings and the distances from

edge to the rainforest interiorsUT ...... 91

TableTU 5.8 Comparisons of regression models of the distribution of seedlings in

different size classes at rainforest edges and the rainforest reference plot..UT ....92

TableTU 5.9 Correlations of the density and the number of species of seedlings with the stem density, basal area and species richness of canopy trees, small trees and

lianas > 2cm at the rainforest edgesUT ...... 93

TableTU 5.10 Correlations between the density of tree seedlings and liana seedlings in

the rainforest reference plot and the rainforest edges.UT ...... 94

TableTU 6.1 Seed densities and number of species in the soil seed bank of the 1 ha xiv

subtropical rainforest plotUT ...... 106

TableTU 6.2 Number of seeds of dominant species germinated from 100 soil samples

within the 1 ha subtropical rainforest plotUT ...... 107

TableTU 6.3 The occurrence of species in the soil seed bank which were also present as

trees or as seedlings within the 1 ha rainforest plot.UT ...... 108

TableTU 6.4 Spearman Rank Correlations between variables from the soil seed bank and those from the above-ground vegetation in 10 m × 10 m plots of the 1 ha

rainforest reference plotUT ...... 109

TableTU 6.5 Moran’s I statistic on the spatial distribution of seeds in the soil seed bank

of the 1 ha subtropical rainforest plotUT ...... 110

TableTU 6.6 Moran’s I and P-value of significance tests from a spatial autocorrelation analysis of the canopy trees of C. paniculosa in the 1 ha subtropical rainforest

pot, based on their x y coordinates within the 1 ha plotUT ...... 110

TableTU 7.1 Summary of germination experiments with dominant plant families according to their contributions to species richness or seed density in the soil

seed banksUT ...... 124

TableTU 7.2 Compositions of the soil seed banks of rainforest edgesUT ...... 125

TableTU 7.3 Numbers of tree seedlings germinated from the soil seed banks from

eucalypt forests and pastures adjacent to rainforest edges.UT...... 126

TableTU 7.4 Comparisons of the number of seedlings germinated from the soil seed banks at different positions from edge to rainforest interiors at different types

of rainforest edges.UT ...... 128

TableTU 7.5 Correlations of the densities of dominant species in the soil seed banks with

the distances from edges to the rainforest interiors.UT ...... 132

TableTU 7.6 Results of the ANOSIM on the composition of soil seed banks among

different type of rainforest edgesUT ...... 134

TableTU 7.7 Spearman Rank Correlations between the seed density and number of

species in the soil seed banks with variables from above-ground vegetationUT 136

TableTU 7.8 Number of canopy trees and number of seeds of species found in both the

soil seed bank and the tree canopies.UT ...... 137

TableTU 7.9 Summary of tree species shared by the soil seed bank and tree canopy at

three types of rainforest edges.UT ...... 138

TableTU 7.10 Occurrence of tree species found in both the soil seed and the seedling

banks at three types of rainforest edgesUT ...... 139

TableTU 7.11 Summary of species found in both the soil seed and the seedling banks at

rainforest edges.UT ...... 140

TableTU 8.1 Classification of tree species according to their occurrence in the soil seed

bank, seedling bank and canopy treesUT ...... 159

xvi

LIST OF APPENDICES

AppendixTU 1 List of plant species and their occurrence in different forms in the present study (tree, liana, seedling and seeds in soil seed bank). Nomenclature and plant life

formsUT ...... 195

AppendixTU 2 The distributions of lianas within liana carrying treesUT ...... 207 2 AppendixTU 3 Distributions of total basal area (mUP /UP ha) of trees (DBH ≥ 5 cm, above)

and lianas (DBH ≥ 2cm, below) within the one hectare rainforest reference plot.UT ..209

xvii

ACKNOWLEDGEMENTS

First of all, I would like to thank my principal supervisor, Roger Kitching, for his support and encouragement through out my study and far beyond. He not only made himself available when things popped up, but also provided me with inspirational guidance and allowed me to share his insightful knowledge of rainforest. To support an international student like myself, Roger had to, not only direct me through the language barriers, but also patiently revised all my chapters during thesis writing,.

I also acknowledge my associate supervisor Jacinta Zalucki. Whenever I got frustrated during the writing up, there was always heart-warming encouragement from her, inspiring me to move on. She also offered me her time revising all my thesis chapters and provided me with many constructive comments.

During my study, Sarah Boulter was more than just a colleague and a good friend, but also an important advisor. She always lent me a hand whenever I needed it, and helped me with my research planning and data collection as well as thesis preparation. Thanks Sarah.

I also want to thank Associate Professor Carla Catterall. Carla treated me as one of her own students and allowed me to join her wildlife discussion group where I learnt the most during my study.

During writing up the thesis, I moved my office down to the wildlife lab, where I spent a lot of time with John Kanowski, a humble, diligent and respected scientist, and a good friend. John provided me with many insightful suggestions about my study and inspired me all the time, especially when I was struggling with my writing. There were many occasions that I directly interrupted into his work for quick answers. John has also critically commented on my thesis.

Much of my knowledge of the local flora came from many discussions and field trips with Stephen McKenna, a fine botanist. Stephen also kindly read some of my chapters when he himself was struggling with his health. I sincerely hope he will get

xviii

his health back soon and carry on his own PhD study.

I also want to thank Cath Moran who kindly provided me her seed dispersal data and commented on two of my chapters. I’m indebted to Scott Piper who provided me with invaluable statistical advice, especially at the time when he was finalizing his own thesis. Thanks also goes to John Holt, the most popular and honoured volunteer in Kitching’s lab, who kindly proof read some of my chapters.

My study involved a large amount of field work which would have just been simply impossible without the help from many volunteers, in particular Michelle Baker, James Bunker, Amanda Norman, Aki Nakamura, Jane Ogilvie, Craig Strong, Kyran Staunton, Brett Taylor, Judy Tian and Mary Wu. Special thanks to Jane who devoted a great deal of her time helping me, and she is great company in the field.

I’m lucky enough to be able to share the international “triangle room” with a group of wildlife enthusiasts, Peter Grimbacher, Terry Reis, Aki Nakamura and Darren Bito. This room is always full of joy, humour, wisdom, and of course friendly “the chicken or the eggs” arguments on many occasions. There are so many good memories which I will be carrying for years to come. In particular, Peter for helping me catch my first ever fish in Australia - unfortunately it was too small to be mounted; Terry for teaching me all my birding knowledge, starting from telling a Butcher Bird from a Magpie; Aki for bravely guiding my first drive to Lamington National Park when I just got my driver’s licence and was still confused about which side of the road I should drive on; and Darren for bringing the office a breath of completely fresh air with his typical PNG sense of humour which made my final writing up so much easier. I remember I was the first one to move into this office, but still haven’t got the chance to use the magic desk, a privilege for whoever is leaving the office. Thanks Aki for moving on, now I am ready to take it over.

Staffs from Griffith University, such as Marianne Mitchell, Dave Henstock, Don Dennis and Bruce Mudway, have helped me on many occasions, smoothing my way towards this completion, in particular, Bob Coutts and Roselyn Steele for organizing and sharing the duty of the bush house maintenance. I also thank my fellow Chinese students in the AES – Zhi-Qun Huang, Jung-Hsun Lien, Chi-ming He and Simon

xix

Song, for supporting each other. Thanks also go to the O’Reilly families for kindly allowing me to work on their properties.

I also want to thank my colleagues and friends from Xishuangbanna Tropical Botanical Garden, the Chinese Academy of Sciences, for their understanding and support, in particular, Prof. Min Cao who first introduced me to rainforest studies. I’m looking forward to being with them again.

The research funding for this study was generously provided by the Rainforest Cooperative Research Centre and the Australian School of Environmental Studies of Griffith University. My study was supported by an International Postgraduate Research Scholarship provided by the Australian Government, a Griffith University Postgraduate Research Scholarship, a Top-up Scholarship provided by the Australian School of Environmental Studies and a Griffith University Completion Assistance Postgraduate Research Scholarship.

Finally, and most of all, I want to thank my dear wife Guangying Zhao, for her understanding, support and love during the past four and half years. Guangying not only patiently took over all the household duties, but also helped me in the rainforest whenever I needed it. After the birth of our daughter, she had to take care of both the baby and the father. We named our daughter Flora, to remember the time we worked together in the rainforest in Lamington National Park, a place described as the last fairyland in the movie “Fern Gully”, a movie Flora picked up by herself and loves the most. I also thank my family in China for their support throughout the course of this study, especially my dad and mum who still haven’t had a chance to see their granddaughter.

xx COLOUR PLATES

Plate 1 Left: An overview of the subtropical rainforest in Lamington National Park (complex notophyll vine forest). Note the emergent tree, Hoop Pine Araucaria cunninghamii, in the far back. Right: Looking into the same type of rainforest. The buttressed tree in the front is Agyrodendron trifoliolatum

Plate 2 Lianas are a conspicuous phenomenon in rainforest but have largely been neglected in rainforest studies. Many large lianas hang free without twisting the trunk of their host trees. A series of smaller size trees may have been used and killed by lianas to access the canopy. Left top: A liana, Melodinus australis, hanging free from the trunk of its host tree, Pseudoweimannia lanchnocarpa. Left bottom: A liana tangle formed by Austrosteenisia glabrastyla possibly fallen from a dead host tree. Right: A new shoot of Celastrus subspicatus starting to climb its host.

xxi

Plate 3 Left: Looking through a roadside rainforest edge. The structure of the rainforest near the edges is kept relatively intact. Right: The high light environment at the roadside has allowed the establishment of many shade- intolerant species such as Pollia crispata (A), Alocasia brisbanensis (B) and Dendrocnide excelsa ( C).

Plate 4 Top: A well fenced pasture/rainforest edge at one of the study sites. Bottom: Many weeds (e.g. Lantana camara, Bottom Left) and rainforest pioneer species (e.g. Solanum aviculare Bottom middle and Solanum mauritianum, Bottom right), can be found near the edge.

xxii

Plate 5 Left: An overview at a eucalypt forest/rainforest edge. The yellow line indicates the location of the edge. Right: Looking inside a eucalypt forest on one of the studied sites. The scorched bark of the eucalypt trees indicate past fire events

Plate 6 Top: Soil samples are spread into plastic trays for germination. Bottom: Seedlings germinated from soil samples (from left to right: Solanum mauritinaum, Phytolacca octandra and Polyscias elegans).

xxiii

1 GENERAL INTRODUCTION

1.1 The Global Rainforest Crisis

The deforestation and fragmentation of rainforest has become a major threat to global biodiversity (Gomez-Pompa et al. 1972; Whitmore & Sayer 1992; Turner 1996; Whitmore 1997; Fahrig 2003; Curran et al. 2004; Wright & Muller-Landau 2006a). According to an assessment of FAO in 1990, tropical rainforest has been lost at a rate 7 of 1.54×10P P ha/ year throughout the 1980s, with an annual deforestation rate of 0.8% (Whitmore 1997). Recent estimates indicate that this rate is still increasing with an annual rate of 1% during the 1990s (FAO 2001). The massive loss of rainforest has not only demolished large areas of habitat, on which the local biota relies, but also subjects the remaining rainforest to edge effects, leading to an on-going long-term degradation. (Whitmore & Sayer 1992; Murcia 1995; Turner 1996; Laurance et al. 1997; Laurance et al. 1998a; Laurance et al. 2000). Many of remaining rainforest are restricted to small, isolated remnants on infertile or unaccessible areas which, nevertheless, act as important refuges for the local biota (Webb & Tracey 1981; Turner & Corlett 1996; Whitmore 1998a). Even ‘protected’ rainforests are continually threatened by illegal extraction of forest products, logging and conversion to agricultural land, due to regional political and economic crises (Nepstad et al. 1999; Curran et al. 2004; Laurance 2004b; Olupot & Chapman 2006). As a result, the area of remaining rainforest is approaching a threshold which may result in a global species extinction crisis (Whitmore 1997; Pimm & Brooks 1999; Sodhi et al. 2004; Wright & Muller-Landau 2006b).

1.2 Rainforest Fragmentation and Edge Effects

A direct result of rainforest fragmentation is the increase in the magnitude of edges, through which generally hostile edge effects may lead to long-term degradation of the remaining rainforest (Janzen 1983; Laurance 1991b; Saunders et al. 1991; Murcia 1995). The edge effects may penetrate considerable distances to the rainforest interior, resulting in changes in species composition, structure and ecosystem processes across the ecotone (Saunders et al. 1991; Murcia 1995; Turner 1996; Mesquita et al. 1999;

Gascon et al. 2000).

Murcia (1995) classified edge effects into (i) abiotic effects and (ii) direct and indirect biological effects, and these edge effects may change in magnitude over time. Most abiotic effects, such as changes in temperature, light regime and moisture, are important and can penetrate deep into forest interiors at newly created edges (Kapos 1989; Williams-Linera 1990b). The abrupt changes in environmental conditions can have profound biological consequences in both the short- and long-terms. One of the direct consequences of these changes is the elevated death of trees, especially large canopy trees, along the edges (Laurance et al. 1997; Laurance et al. 2000). The changed microclimate conditions in the forest, especially the increase in light penetration, may lead to a sudden release of the existing seedling banks (Williams- Linera 1990a; Sizer & Tanner 1999), and cause germination or even direct death of seeds in soil seed banks(Williams-Linera 1990b). These short term changes, over time, may have profound impact on the species composition, structure and dynamics of the rainforest edges (Gascon et al. 2000; Laurance et al. 2006a). As the edges age, many abiotic effects tend to weaken progressively or become even undetectable after the closure of the edges by rainforest pioneer species and lianas (Williams-Linera 1990a; Turton & Duff 1992; Kapos et al. 1997; Turton & Freiburger 1997; Williams-Linera et al. 1998). The altered forest composition and structure near rainforest edges, with the interaction with the surrounding matrices, however, may generate long-term direct and in-direct biological edge effects (Saunders et al. 1991; Murcia 1995; Laurance 2004a).

The most noticeable changes at rainforest edges are the dramatic increase in abundance of early successional tree species (Williams-Linera 1990a; Laurance 1991a; Oliveira-Filho et al. 1997; Laurance et al. 1998b; Laurance et al. 2006b), and the proliferation of lianas (Caballe 1984; Laurance 1997; Oliveira-Filho et al. 1997; Laurance et al. 2001b). This may lead to a secondary succession in rainforest near the edges, which, to some extent, is similar to gap dynamics within rainforest interiors after natural disturbances (Denslow 1987; Williams-Linera 1990b; Laurance et al. 1998a). The increase in the abundance of early successional species may cause significant changes in species composition, structure and even modify the regeneration patterns of rainforests near the edges (Oliveira-Filho et al. 1997;

Laurance et al. 1998a; Laurance et al. 1998b). Nonetheless, these changes in forest composition and structure may help ‘seal’ the rainforest edges, and thus reduce the penetration of many biotic and abiotic edge effects into rainforest interiors (Janzen 1983; Williams-Linera 1990b; Cadenasso & Pickett 2000).

Apart from the changes in plants, the changes in vertebrates, invertebrates, and microorganisms may also have profound impact on the dynamics of rainforest edges (Turner 1996; Laurance 2000; Terborgh et al. 2001; Fahrig 2003). Changes in animal assemblages may alter many ecological processes such as pollination (Aizen & Feinsinger 1994; Murren 2002), seed dispersal (Wunderle 1997; Holl 1998), seed predation (Nepstad et al. 1996) and herbivory (Wahungu et al. 1999; Cadenasso & Pickett 2000) which, in turn, affects the vegetation dynamics near the edges. Currently, a lack of seed dispersal agents, especially large sized vertebrates that disperse large rainforest seeds, has become a threat to the regeneration of many rainforest fragments (Wright et al. 2000; Terborgh et al. 2001). In addition, the removal of large predators may have a cascading effect on other trophic levels. This could lead to a dramatic increase in herbivores and seed predators in remnant rainforests causing increases in seed predation and damage to seedlings (Terborgh et al. 2001). Changes in microorganisms, such as pathogens, can cause elevated levels of seedling infection (Benítez-Malvido & Lemus-Albor 2006) and high death rates of seeds in the soil seed bank (Garwood 1989). Moreover, many of above changes may also interact with each other (Murcia 1995). For example, increases in herbivore damage may promote the incidence of pathogen infection at rainforest edges (Benitez-Malvido & Lemus-Albor 2005).

Changes in the microclimate and forest composition and structure may also modify ecological processes that operate as a result of the interactions between biotic and abiotic edge effects (Murcia 1995). Leaf-litter decomposition, for example, may be altered at rainforest edges through the changes in microclimatic variables, such as temperature and moisture and, hence, affect the activities of leaf-litter invertebrates (Didham 1998). An increase in fire events at rainforest edges may be induced by changes in microclimate, forest composition and structure, representing a synergistic interaction between edge effects and human disturbances (Cochrane & Laurance 2002). These changes in ecological processes may have prolonged effects on the

regeneration of rainforest remnants.

Apart from the changes at the rainforest edges per se, the species composition and structure of surrounding matrices can have profound impacts in determining the magnitude and distance of edge effects (Janzen 1983; Saunders et al. 1991; Didham & Lawton 1999; Mesquita et al. 1999; Harper et al. 2005; Denyer et al. 2006). The understanding of the movement of species between rainforest edges and surrounding matrices can have important applications in rainforest conservation as well as for the restoration of deforested areas (Janzen 1983; Aide et al. 2000; Holl et al. 2000) (Turner & Corlett 1996).

Edge effects can also have a large scale impact on the microclimate in fragmented landscapes (Chen et al. 1999), and alter ecological processes from edges deep into rainforests (Laurance 2000). This, in turn, will lead to a long-term degredation of the remining rainforests and may largely compromise their conservation value.

Although there is a paucity of direct evidence, several long-term rainforest monitoring studies have suggested a possible global response of rainforests to the increase in

atmospheric CO2B B concentration and air temperature. These responses include an increase in above-ground biomass and dominance of lianas, an acceleration of forest turnover rate, and a possible shift in long term tree regeneration patterns (Phillips et al. 2002; Laurance 2004a; Lewis et al. 2004; Wright et al. 2004). Climatic changes, such as an increase in cyclone occurrence and/ or intensity and rainfall seasonality, may interfere substantially with rainforest regeneration by modifying the rainforest seedling banks (Whitmore 1998b). Moreover, these changes are likely to be amplified by an interaction between climate change and rainforest fragmentation (Laurance et al. 1998c; Laurance 2004a; Opdam & Wascher 2004).

1.3 The Future of Rainforest Remnants

Ecologists have long realized that the rainforest is a nonrenewable resource (Gomez- Pompa et al. 1972). This, however, did not stop the massive deforestation of recent decades, driven mainly by the increase in human population (Whitmore 1997; Tilman et al. 2001; Wright & Muller-Landau 2006b). It is clear that rainforest deforestation

and subsequent fragmentation will still be the main threats to the global biodiversity for years to come (Whitmore 1997; Laurance 2007). Meanwhile, the remaining rainforests have to face new emerging threats from rainforest fragmentation and interactions with climate change (Wright 2005; Laurance & Peres 2006; Lewis 2006).

At the same time, tremendous efforts have been made internationally to mitigate the rainforest crisis by establishing and managing protected areas (James et al. 1999; Margules & Pressey 2000; Bruner et al. 2001; Fagan et al. 2006) and restoring at least part of the deforested areas (Lamb et al. 2005). Maintaining large areas of intact primary rainforest is still the most efficient mechanism of rainforest conservation (Turner & Corlett 1996; Bruner et al. 2001; Laurance & Peres 2006). Restoration of deforested areas and the management of degraded forests, however, are also of great importance (Lamb et al. 2005; Wright 2005). Recently, Wright and Muller-Landau (2006) predicted a global decrease in deforestation rate and an increase in natural forest regeneration in the near future, based on a projected slowing down of global population growth and the intensification in urbanization. This scenario has highlighted the importance of natural regeneration of deforested areas (Laurance 2007). The rainforest remnants may play important roles in promoting natural regeneration, by acting as species pools for the local biota (Turner & Corlett 1996; Aide et al. 2000; Lamb et al. 2005).

1.4 Rainforest Fragmentation: an Australian Perspective

Covering less than 0.5% of the continental land area (Cofinas & Creighton 2001), the rainforests of Australia are mainly confined to the east coastal areas and extend from northern Australia to Tasmania (Webb & Tracey 1981). Despite being small in area, the rainforests support a significant proportion of Australia’s biodiversity and have been important biological refugia during geological times (Webb & Tracey 1981; Schneider & Moritz 1999). These refugial rainforests are typically isolated patches within large tracts of sclerophyllous vegetation; these two vegetation types are completely different in terms of their species composition and structure(Webb 1968; Adam 1992; Bowman 2000).

1.4.1 Rainforest fragmentation in a geological context

During the early Miocene rainforests once occupied a large area of Australia; but with an increasingly arid climate towards the end of the Tertiary, rainforests contracted to the moister continental margins, and became severely fragmented due to repeated contractions during the glacial events of the Pleistocene (Adam 1992; Kershaw et al. 2002). These cooler and drier conditions favoured sclerophyll vegetation which expanded enormously at the expense of rainforest (Singh et al. 1981; Kershaw 1994; Kershaw et al. 2002). Evidences from pollen and charcoal from sedimentary studies (Kershaw 1978, 1981; Singh et al. 1981; Kershaw 1994) and more recent charcoal dating (Hopkins et al. 1990b; Hopkins et al. 1993; Hopkins et al. 1996) have found that fire has played an important role in the seesaw battle between rainforest and sclerophyllous vegetation. The expansion of eucalpyt woodland and contraction of rainforest continued up to about 8000 yr BP when rainforest started to expand following a shift of climate to warmer and wetter conditions (Ash 1988; Hopkins et al. 1993; Hopkins et al. 1996; Kershaw et al. 2002). This rainforest re-expansion is still ongoing in many rainforest areas, influenced by changes in local fire regimes (Smith & Guyer 1983; Ash 1988; Unwin 1989; Harrington & Sanderson 1994; Russell-Smith et al. 2004a).

1.4.2 Aborigines and rainforest fragmentation

Aboriginal people arrived in Australia about 56 ka years ago when the low sea level allowed access by a possible land-bridges and short voyages from Southeast Asia (Roberts et al. 2001; Clarke & Clarke 2002). Early hunting and gathering activities by Australian Aborigines seems to have had only limited impact on the rainforest (Bailey et al. 1989; Richards 1996). As the population increased these impacts are likely to have increased and might have changed the already fragmented Australian rainforest substantially (Cosgrove 1996). The most destructive disturbance would have been the use of fire by aboriginal people as a hunting tool which has largely modified the Australian landscape and has become a primary force regulating the rainforest/ sclerophyll forest edges (Smith & Guyer 1983; Kershaw 1986; Ash 1988; Unwin 1989; Russell-Smith et al. 2004b). This practice was well documented by early European explorers (Fensham 1997, Preece 2002) and is supported by evidence from the pollen and charcoal record (Lesley 1989; Hopkins et al. 1993; Kershaw 1994;

Hopkins et al. 1996; Kershaw et al. 2002). The use of fire, along with climatic changes, was largely responsible for the rainforest decline during the late Pleistocene (Kershaw 1986; Hopkins et al. 1993; Kershaw 1994; Hopkins et al. 1996).

Compared with other rainforest areas, one of the most significant differences of rainforest exploitation in Australia is that the aboriginal people never practiced agriculture (Cosgrove 1996; Richards 1996; Bowman 2000). Long term shifting agriculture has had a profound impact on the composition and dynamics of rainforest elsewhere and is largely responsible for rainforest deforestation and fragmentation (Myers 1993, 1994; Whitmore 1997; Willis et al. 2004). The large area of secondary forest created during shifting agriculture may potentially regenerate into forest similar in composition and structure to the original rainforest through natural processes. These secondary rainforests can have great conservation value in mitigating the pressure from current deforestation (Myers 1994; Richards 1996; Chokkalingam & De Jong 2001; Wright 2005). On the other hand, secondary forests often contain agricultural weeds which can interfere with the dynamics of the rainforest, especially along the edges. This may lead to a long term degradation of rainforest remnants (Janzen 1983; Laurance et al. 2002; Nascimento et al. 2006). In Australia, aboriginal people use fire for hunting mainly in areas outside rainforest (Floyd 1990; Adam 1992; Fensham 1997). No large area of secondary rainforest forest would have been generated after burning as the burnt areas are generally claimed by sclerophyllous species that are well adapted to fire (Gill 1981; Floyd 1990). Even so, fires may frequently scorch rainforest near the eucalypt forest/ rainforest boundary and act as an important factor in regulating the dynamics of the rainforest/ open forest edges, which are generally very sharp (Smith & Guyer 1983; Kershaw 1986; Ash 1988; Unwin 1989; Russell-Smith & Stanton 2002; Russell-Smith et al. 2004b). In the absence of fire, rainforest may gradually push into sclerophyllous forest at a very slow rate, for example 12 m over 10 years in rainforest/ tall-open forest edge (Unwin 1989). Several studies have found progressive rainforest expansion after the reduction of Aboriginal fire practices (Harrington & Sanderson 1994; Russell-Smith et al. 2004a, but see hill, 2000).

1.4.3 European impacts on rainforest

During the past 200 years, European impacts have become the driving force behind changes to Australian rainforest following the cessation of Aboriginal influences (Hobbs & Hopkins 1990; Hill et al. 2000). The rainforests, especially those in the lowlands and on high nutrients soils, have been extensively cleared and converted to agricultural uses or developed for dwellings (Hobbs & Hopkins 1990; Adam 1992; Glanznig 1995). According to the Australian Native vegetation Assessment (2001), over 30 percent of pre-European rainforest (including vine thickets) has been cleared, with most lowland areas having lost almost all of the original rainforest. The most extensive rainforest destruction has occurred in the subtropical zone, where more than half of the original rainforest has been cleared (Kanowski et al. 2003). For example, the ‘Big Scrub’ rainforest area in northern New South Wale was once covered with 75,000 ha of extensive subtropical rainforest before European settlement; now, the rainforest has been reduced to just 300 ha in 40 isolated remnants (Floyd 1987). Many of the surviving rainforests have been, more or less, logged and have been subjected to long term degradation (Horne & Hickey 1991). Apart from clearing and logging, the remaining rainforests have also been subjected tolong term edge effects associated with agricultural activities in adjacent areas (Hopkins & Graham 1984; Graham & Hopkins 1990; Hopkins et al. 1990a; Laurance 1997; Hill et al. 2000). Many of the existing remnants have been subjected to internal fragmentation caused by road construction and the creation of power line corridors (Goosem 1997; Pohlman et al. 2007) and threatened by invasive species, such as feral pigs (Pavlov et al. 1992; Laurance & Harrington 1997), exotic vines and scramblers such as lantana (Lantana camara) (Gentle & Duggin 1997) and many other species (Goosem & Turton 2006).

1.4.4 Australian rainforest in a changing climate

Due to its evolutionary background, long term history of fragmentation and unique disturbance regimes, Australian rainforest may be much more vulnerable to climate change than rainforest in other areas (Hilbert et al. 2001; Williams 2003; Williams & Hilbert 2006). First, the highly endemic nature of Australian rainforests implies that any loss of rainforest area may result in the loss of many species (Webb & Tracey 1981; Crisp et al. 2001; Williams 2003). Second, the highly fragmented rainforest landscape, produced both by geological and anthropogenic factors, may largely

restrict the ability of species to respond to the climate change (Hilbert et al. 2001; Williams 2003). The surrounding sclerophyll flora is so contrasting in species composition, structure and key ecological processes (particularly fire) to the adjacent rainforest that it may become an obstacle limiting the movement of many species among rainforest remnants. Third, most of the remaining rainforests are confined within the coastal area which may be subjected to projected increases in cyclone intensity under a climate change scenario (Walsh & Ryan 2000). These impacts may also interact with edge effects causing considerable damage to rainforest remnants (Laurance 1997; Laurance 2004a). In addition, an increase in fire events and fire

intensity has been predicted in Australia based on the increase in atmospheric CO2B B concentration (Williams et al. 2001; Cary 2002). This may pose a great threat to the rainforest by changing the dynamics between rainforest and sclerophyll vegetations and may compromise the potential for rainforest expansion (Hopkins et al. 1996; Russell-Smith et al. 2004b).

1.5 Rainforest Edge Studies in Australia

Fragmentation has become a significant feature of the Australian rainforest (Webb & Tracey 1981). An understanding of the vegetation dynamics across the edges is important for the conservation of rainforest as well as for the restoration of adjacent deforested areas. Many of the rainforest remnants act not only as a refuge for the remaining biota but are also an important species pool for the recovery of degraded areas (Turner & Corlett 1996; Lamb et al. 2005).

Rainforest edges have received considerable attention in rainforest studies in Australia. The studies, however, have been primarily focused on rainforest/ sclerophyll forest edges and concentrated mainly on the composition, structure and the dynamics of the ecotone and the relationships with local fire regimes.(Jackson 1978; Smith & Guyer 1983; Ash 1988; Duff & Stocker 1989; Unwin 1989; Turton & Duff 1992; Wang 1997; Russell-Smith et al. 2004b). The rainforest / wet sclerophyll forest edges, in particular, have received great attention as these unique habitats support many endemic small mammals (Webb & Tracey 1981; Ash 1988; Harrington & Sanderson 1994; Williams & Marsh 1998). Few of those studies, however, have discussed rainforest edges in a fragmentation context (Fox et al. 1997). Other anthropogenic edges have received

very little attention (but see Willson & Crome 1989; Laurance 1991a; Hester & Hobbs 1992; Scougall et al. 1993; Fox et al. 1997; Goosem 1997; Laurance 1997; Turton & Freiburger 1997; Pohlman et al. 2007).

Most of the studies cited above have focused exclusively on one type of edges without taking the surrounding matrices and associated disturbances into consideration. These studies are concentrated mainly on the composition and structure of edges and the distribution of weeds across the ecotone. Other important regeneration components, such as the soil seed bank and seedling bank, have been largely neglected in edge studies (but see Willson & Crome 1989; Wang 1997; Tang et al. 2003).

Recent studies have suggested that lianas may play an important role in the dynamics of rainforest edges (Tabanez et al. 1997; Laurance et al. 2001b; Schnitzer & Bongers 2002). Although some studies have discussed briefly the edge-related changes in liana assemblages in tropical and subtropical Australian rainforest (Hegarty 1989; Laurance 1991b, 1997), so far there has been no systematic study of the distribution of lianas in Australian rainforest edges. The lack of such information may limit our understanding of ecological processes that regulate Australian rainforest edges and hamper the development and implementation of proper management plans.

1.6 Objectives of This Thesis

The aim of the research described in this thesis is to contribute to the understanding of the dynamics of rainforest edges through the comparison of the species composition and structure of the tree canopy and associated soil seed banks, seedling banks and lianas under different disturbance regimes. The main focus is to describe the distribution patterns of rainforest plants from the edge to the rainforest interior and to compare these distribution patterns among different types of edges. Meanwhile, the spatial distribution and associations of rainforest trees, lianas, seedlings and seeds in the soil seed bank within a 1 ha undisturbed rainforest reference plot has also been conducted to assist the understanding of the responses of rainforest plants to edges. The results of this study may help in assessing the current situation and future dynamics of rainforest edges and to identify management priorities.

1.7 Structure of This Thesis

The structure of this thesis follows a top-down sequence, from canopy trees and lianas, through seedlings, down to the soil seed bank. Chapter 2 introduces the study site and gives an overview of the study design and methods. A brief introduction to subtropical rainforest and its regeneration is also included. Chapter 3 focuses mainly on a comparison of the composition and structure of the rainforest tree canopy among three types of rainforest edges as well as within the 1 ha rainforest reference plot. The distribution patterns of rainforest trees in response to edge effects are also assessed with special consideration of different plant regeneration strategies. Chapter 4 describes the distribution of lianas from rainforest edges to interiors, and their responses to different types of rainforest edges compared, again, with the rainforest reference plot, focusing on the changes in liana density, life form and climbing mechanisms. The association between lianas and trees and their interaction with edge effects is also discussed. Chapter 5 deals with the distribution of rainforest seedlings, including both tree and liana seedlings, at different rainforest edges and the rainforest reference plot. This chapter assesses how the species composition and structure of the forest canopy, in conjunction with various disturbances affect the composition and distribution of seedlings in the seedling banks. Chapter 6 examines the composition and spatial distribution of the soil seed banks and their relationship with the above- ground vegetation, including trees, lianas and seedlings in the rainforest reference plot. The relationship between viable seeds in the soil seed bank and the spatial distribution of conspecific canopy trees is discussed for some dominant species, testing the dispersal limitation hypothesis. Chapter 7 compares the composition and distribution of soil seed banks from the edge to the rainforest interior across different types of rainforest edges. This chapter concentrates on the relationship between the soil seed banks and the above ground vegetation including trees, lianas and seedlings in order to test the function of edges in maintaining rainforest soil seed banks. By comparing the existing soil seed banks, this chapter also describes the movement of seeds between rainforest and surrounding matrices, and considers the implications of these results for in rainforest restoration. Finally, chapter 8 brings together the main findings of this study and discusses future research directions. The implications of this study for rainforest conservation are also discussed.

2 STUDY AREA AND METHODOLOGY

2.1 Locality and Geography

The research conducted for this thesis was located in, or immediately adjacent to, Lamington National Park, about 100 km south of Brisbane, near the Queensland/ New South Wales border (28°13´S, 153°07´N , Fig. 2.1). The area forms part of the McPherson Range, running north to north-west along from the caldera rim of the ancient Mt. Warning shield volcano (Stevens 1977). The altitudes of the National Park range from 350 m to about 1200 m above sea level. The soils are mainly basaltic derived from Cainozoic igneous rocks (Stevens 1977). This National Parkm, with a total area of 20, 600 hectares, preserves the part of the largest subtropical rainforest remnant in Australia and is a major component of the World Heritage Area, Gondwana Rainforests of Australia (formerly “Central Eastern Rainforest Reserves of Australia”), declared in 1995.

2.2 Climate

The climate of Lamington National Park area is a typically moist and subtropical, with a distinct summer-autumn wet season and a cool winter (Fig. 2.2). The highest temperature may reach 35°C during summer (December) and the lowest can approach 0 °C during winter (in June).

This area receives an average annual rainfall of 1601mm (Australian Bureau of Meteorology), ranging from 707 mm in 2002 (two years before the present study), the driest year ever recorded, to 3768 mm in 1974 (Fig. 2.3). In recent years, the area has been subject to a prolonged drought, with annual rainfall below average since 2000.

Lahey’s Memorial

Bristle Bird Creek

One hectare plot

Luke’sLuck’s FarmFarm

Patts Bluff

Figure 2.1 The location of Lamington National Park and the study area ( Eucalypt forest/ rainforest edges Pasture/ rainforest edges Roadside rainforest edges)

250 30 ( average temperature Monthly

25 200

20 150 15 100 10

Monthly rainfall (mm) (mm) Monthly rainfall 5 ˚ C) 0 0 123456789101112

Figure 2.2 The monthly rainfall (bars) and average temperature (squares with dashed line) of the study area. Rainfall data obtained from Green Mountains, national weather station number 40182 at 28°13'52"S and 153°08'08"E, altitude 917.0m, averaged over 77 years; Temperature data obtained from Mt Tamborine, national weather station number 40197 at 27°58'10"S and 153°11'43"E, altitude 515.0m, averaged 41 years of records. The altitude of the study area is around 350 m higher than Mt Tamborine weather station thus the average temperature should be around 1.8 degree lower than that shown in the figure. Data source: Bureau of Meteorology, Australia Government

4000 1974

3500

3000 1988 2500 1999 2000

1500

1000 1957 500 2002 0

Figure 2.3 Long term patterns of the average rainfall of the study area (heavy dashed line shows the average annual rainfall. Records obtained from Green Mountain, national weather station number 40182 at 28°13'52"S, 153°08'08"E,altitude 917.0m, with available data from 1916 to 2006. Data source: Bureau of Meteorology, Australia Government.

2.3 Vegetation

Lamington National Park is located in a region where elements of the northern and southern Australian biota can be found (the 'McPherson - Macleay' botanical overlap - Burbidge 1960), which supports both tropical and temperate species (Hopkins et al. 1977). It has also been identified as one of the centers of endemism for Australian biodiversity (Crisp et al. 2001).

The floristic features of the Park have been detailed by McDonald and Thomas (1989). This park contains about 880 species of vascular plants of which 61 species have been nominated as endangered, vulnerable or rare (McDonald & Thomas 1989). The rainforests in the park can be broadly classified as warm subtropical, cool subtropical, warm temperate or cool temperate rainforest following a gradient from lower to higher altitudes (McDonald & Thomas 1990). At lower altitudes, drier forms of rainforest characterized by emergent hoop pine (Araucaria cunninghamii) can also be recognized (McDonald & Thomas 1990). Apart from rainforest, sclerophyll vegetation is also common within and around the National Park, varying in forms from wet sclerophyll forest to mountain heath (Graham et al. 1977). The most common form of sclerophyll vegetation is the open forest dominated by the genus Eucalyptus, with a grassy understory on basaltic soil. Extensive ecological studies have been conducted on the composition, dynamics and regeneration of the rainforests within the National Park (Hopkins 1975; Connell et al. 1984; Olsen & Lamb 1988; Abdulhadi 1989; Laidlaw et al. 2000; Kitching et al. 2005).

2.4 The History of Lamington National Park

Aboriginal people known as the Yugambeh language group had inhabited the Gold Coast and hinterland areas long before European settlement. Wangerriburra, Birrunburra, Kombumerri and Migunberri tribes were the main residents around today’s Lamington National Park area (Jarrott 1977; Ogilvie 2006). Archaeological remains found at Bushranger Cave show evidence of Aboriginal hunting activities in the present rainforest area at least 6, 000 years ago (Hall 1986). Although there is no direct record, it is likely that the Aboriginal people have had significant impact on the rainforest by long-term extraction of forest products. In addition, their burning of

open forests to assist hunting, a common Aboriginal practise in Australia (Bowman 2000), might also have shaped the rainforest landscape especially near the rainforest/ open forest edges around the National Park.

The most intensive rainforest destruction in this area happened after European settlement around 1840 (Jarrott 1977). Since then, extensive rainforest clearing has been carried out mostly on fertile lowland areas for dairy, cattle farming and other agricultural uses (Douglas 1977). The deforestation was so devastating that some areas eventually lost almost all the original rainforest (Floyd 1990). Before the arrival of Europeans, the “Big Scrub” area of northern New South Wales was once covered by 75, 000 ha of subtropical rainforest, but now has only 300 ha in about 40 isolated remnants (Floyd 1990). Due to difficulties of access at that time, the area where Lamington National Park lies today was relatively untouched, but early cedar collectors might well have penetrated deep into the rainforest by following stream systems (Jarrott 1977).

The establishment of Lamington National Park followed a long campaign led by Robert Collins and his successor Romeo Lahey from the late 1880s until 1915 when the National Park was officially declared (Jarrott 1977). Following the construction of Lamington National Park road, which began in 1926 and was sealed progressively, a large amount of rainforest timber, especially hoop pine Araucaria cunninghamii, was extracted from areas surrounding the current national park, such as the Mountain Cainbable area (Jackson & Stephenson 1986). Meanwhile, a track system has been built progressively for tourist access. The main body of the track system was completed during the 1930’s (Ogilvie 2006) and the National Park has now become one of the most popular international tourism destinations in Australia, with more than 500,000 visitors per year from all over the world (Weaver 2002). It is also one of the most attractive places for naturalists and scientific researchers due to its unique geological history, its diverse habitats and associated wildlife, and its ready accessibility (Laidlaw et al. 2000).

A key event in the history of this park was the settlement of the O’Reilly family within what became the core area of the park. Following the free selection of land in 1911, just a few years before the establishment of the National Park, extensive

rainforest clearing and burning was carried out to establish dairy farms around the area of the current guest house (O'Reilly 1944). Subsequently, large tracts of secondary forest have been generated after the abandonment of some part of the clearings (Hopkins 1975). This has shaped the local landscape substantially and may have had a profound impact on the regeneration of the rainforest (Hopkins 1975; Abdulhadi 1989).

2.5 Study Area and Sites

The study sites were spread out within an area extending north to Pine Mountain, south to near the O’Reilly’s Guest House, west along a direction from Luke’s farm, Duck Creek Road, to Bristle Bird Creek following the distribution of rainforest, and east near The Lamington National Park Road (Fig. 2.1). The extent of the total study area was about 7 km (North to South) × 4 km (East to West).

Study sites include a 1-ha rainforest reference plot and three types of rainforest edges (Table 2.1). To focus on edge effects, independent of habitat loss, the edges in this study are all connected to continuous rainforest, with only one side facing the surrounding matrices. The eucalypt forest/ rainforest edges were also selected on the base of no signs of recent fire or other major disturbance, such as cyclones or logging. Rainforest/ pasture edges were chosen with pasture directly connected to continuous rainforest and were well fenced without obvious cattle disturbance. Road edges were chosen away from recent tree falls as rainforest gaps may confuse the edge effects.

2.5.1 Reference rainforest

The 1 ha rainforest reference plot was established in 1995 as a permanent biodiversity monitoring plot (Kitching et al. 2005). The plot is located on an east facing slope approximately 1.5 km northwest of O’ Reilly’s Guest House (Fig. 2.1). The 1 ha rainforest plot was subdivided into one hundred 10×10 m grid squares and the corners of the squares are permanently pegged out. The tree flora here has been classified as a complex notophyll vine forest within an Argyrodendron trifoliolatum - Argyrodendron actinophyllum-Caldcluvia paniculosa tall closed-forest alliance which consists of elements from two floristic associations (Laidlaw et al. 2000). The Argyrodendron spp

association is generally distributed below 800 m altitude and is also classified as warm subtropical rainforest. The C. paniculosa - Cryptocarya erythroxylon - Dysoxylum fraseranum - Geissois benthamii association is mainly confined to altitudes from 800-1000 m and is also called cool subtropical rainforest (McDonald and Whiteman 1979). The occurrence of species from the two associations has made this 1 ha plot very diverse in terms of its tree flora when compared with other subtropical rainforest plots (Laidlaw et al. 2000).

2 The initial survey (in 1995) recorded 74 tree species with a total basal area of 70.5 mP P from 1,226 trees above 5 cm DBH (Laidlaw et al. 2000). The dominant tree species, in terms of importance value index, are A. trifoliolatum, Atractocarpus benthamianus, Actephila lindleyi, Baloghia inophylla and C. paniculosa (Laidlaw et al. 2000). The forest has an uneven canopy with emergent trees such as Araucaria cunninghamii and Ficus watkinsiana, extending above 40m. Many canopy trees form distinctive plank buttresses. Robust lianas are conspicuous and dominated by Melodinus australis and Austrosteenisia glabristyla while the epiphyte Pothos longipes is common on many tree trunks. The walking stick palm Linospadix monostachya is very abundant in this plot. The ground is covered mainly by rainforest tree seedlings and ferns such as Lastreopsis spp and Asplenium spp.

2.5.2 Rainforest/ eucalypt forest edge

Rainforest/ eucalypt forest edges were distributed along the western boundary of the National Park from Bristle Bird Creek, Duck Creek, to the south west of Luke’s farm altitudes ranging from 700m to 800m. The edges between rainforest and open eucalypt forest were all abrupt. The rainforests were mainly complex notophyll vine forest (CNVF,Webb et al. 1984) within the Argyrodendron spp association (McDonald & Whiteman 1979). Rainforest around Bristle Bird Creek is slightly drier showing a distinctive emergent layer of A. cunninghamii. This type of rainforest is classified as Araucarian notophyll vine forest (Webb & Tracey 1981), a drier form of the Argyrodendron spp. association with broadly similar composition to the typical association (McDonald & Whiteman 1979).

20 Table 2.1 Summary of sampling efforts for rainforest edges and the rainforest reference plot (Sampling strategy see Fig. 2.4). Vegetation category and Rainforest edges Reference rainforest Eucalyptus forest Pasture sample size Tree (10×10 m plot) 135 (5 distances × 9 transects 100 27 - ×3 edge types) (5 distances × 9 transects) Liana (10×10 m plot) 135(5 distances × 9 transects 100 27 - ×3 edge types) (5 distances × 9 transects) Seedling (1×1 m plot) 1125 (5 subplots×5distances 250 135 - × 9 transects ×3 edge types) (5subplots×3distances × 9 transects Soil seed bank 270 (2 samples× 5 distances 100 54 54 (10×10×5 cm soil sample) × 9 transects ×3 edge types) (2 samples× 3 distances × 9 transects)

The open forests are dominated by the genus Eucalyptus with a grassy understory with slightly different compositions from place to place (McDonald & Thomas 1990). The open forest around Bristle Bird Creek is dominated by Eucalyptus biturbinata and E. acmenoides with occasional Allocasuarina littoralis and a grassy understory. Some rainforest species such as Alphitonia excelsa, A.cunninghamii, Diospyros pentamera and Polyscias elegans form a sparse shrub layer.

The open forest near Duck Creek Road is dominated by Eucalyptus microcorys with a grassy understory. Other Eucalytus species such as E. notablis, E. biturbinata and E. acmenoides can also be found as large canopy trees. This open forest has only one tree layer formed by the large eucalyptus trees, with black scorched bark indicating past fire events. Occasionally, some rainforest species, mainly P. elegans and Guioa semiglauca, can be found as small trees.

The open forest near Pat’s Bluff is dominated by E. microcorys, Eucalyptus campanulata and Eucalyptus quadrangulata. A dense rainforest understory has formed mainly by early successional species such as A. cunninghamii, Dendrocnide excelsa, D. fraserianum, P. elegans and Synoum glandulosum. Some primary rainforest species, such as A. lindleyi, A. actinophyllum, A. trifoliolatum, D. pentamera and Pseudoweinmannia lachnocarpa are already well established. This type of open forest is considered a transition between closed rainforest and grassy open forest, a result of fire exclusion (McDonald & Whiteman 1979).

To sample the eucalypt forest/ rainforest edge, nine 150 m transects were established, each of which stretched 100 m from the edge into rainforest and 50 m from the edge to eucalypt forest. Three transects were near Bristle Bird Creek, three were near Duck Creek Road and the other three transects were close to Pat’s Bluff (Fig1).

2.5.3 Rainforest/ pasture edge

Rainforest/ pasture edges were located mainly around Luke’s Farm and the Pine Mountain Farm near the Lahey Memorial. Their altitudes ranged from 700 m to about 850 m asl. The rainforests are complex notophyll vine forest (Webb et al. 1984) within the tall closed forest of the Argyrodendron spp. association (McDonald & Whiteman 1979). The edges were created at various times but most of them are more than 50

years old (Hopkins 1975). All the edges have been well fenced, preventing cattle entering into the rainforest. A strip of dense shrubs has developed along the fence, which has, more or less, helped “seal” the edges. The pastures are still in use for cattle farming.

To sample the pasture/ rainforest edge, nine 150 m transects were established which stretched 100 m from the edge into rainforest and 50 m from the edge into pasture. Three transects were set up at the edges near Pine Mountain cattle farm and the other six transects were all located at Luke’s Farm (Fig. 2.1).

2.5.4 Roadside rainforest edge

Road edges were distributed along the Lamington National Park Road, starting from about 100 m past the Lahey Memorial towards O’Reilly’s Guest House. This part of the road was widened progressively from a horse track (Jackson & Stephenson 1986). No large machinery was used during the construction of this part of the road and the removal of large trees was deliberately minimized to avoid heavy disturbance. The altitude of the selected edges ranged from 750 m to about 900 m. The forests were complex notophyll vine forest mainly within the Agyrodendron spp association except for the three edges around 900 m altitude which had a similar species composition to the reference rainforest characterized by the occurrence of C. paniculosa.

To sample the roadside rainforest edges, nine 100 m transects were established which stretched 100 m from roadside into the rainforest. The selection of transects followed 9 randomly generated distances starting from the Lahey Memorial.

2.6 Vegetation Survey

2.6.1 Edge vegetation survey

Along each transect, five 10 m ×10 m plots were established at about 20 meter intervals from the edge to the rainforest interior. Around the rainforest/ eucalypt forest edges, three plots of the same size (10 m×10 m) and spacing were also established within the eucalypt forest along each transect (Fig. 2.4). Vegetation surveys were based on the standardized 10 m ×10 m plots in which trees, lianas, and seedlings were

identified and recorded in the whole plot or using subplot (in the case of seedlings).

Tree All tree stems with a DBH above 1 cm were identified and measured at 1.3m height for diameter. Calipers were used to measure stems below 5 cm in DBH. A circumference tape was used for stems 5 cm in DBH and above. Multiple stems that separated below 1.3 m were considered as independent individuals. For trees with buttresses, the measurements were taken above the buttress where the trunk became rounded.

Seedling “Seedlings” in this study were defined broadly as trees or self-supporting lianas that are below 2.5 m in height with at least one true leaf. Herbaceous plants were not included nor were the re-sprout shoots of lianas. When necessary, the seedling root was excavated to check if it was independently rooted.

To sample seedlings, five 1 m × 1 m seedling plots were marked out, with one at each of the four corners and one at the centre of each 10 m ×10 m tree plot. In each subplot, all seedlings were identified and their height measured. For unidentified species, specimens were collected and a search for larger sized individuals was also carried out to assist identification.

Lianas Lianas in this study were defined as woody vines that need physical support to access the canopy. The method used for the liana survey was similar to a proposed standard protocol (Gerwing et al. 2006). Hemiepiphytes, such as strangler figs, are not included in this study. No rattans or climbing Poaceae were recorded in this study. The minimum stem diameter included in the survey was 2 cm.

Liana crowns are important in maintaining the structure of the forest canopy, which may help with the closure of rainforest edges (Janzen 1983; Putz 1984b; Putz 2004).This study focused mainly on the comparison of edge composition and structure and its function in maintaining the soil seed and seedling bank. Only lianas with most of their crowns within the tree plot were included in the survey regardless

whether or not the rooting position was in the plot. Lianas that rooted in the plot but ascended out of the plot were not considered. This criterion is different from that suggested by Gerwing et al. (Gerwing et al. 2006) where they recommended that only lianas rooted within the plot be included in the survey.

Liana diameter was measured at 130 cm above ground level. Cylindrical liana stems were measured using calipers for stems < 5 cm (Diameter) and a tape measure for stems ≥ 5 cm (Circumference). For non-cylindrical liana stems, such as Tetrastigma nitens, the measurements were taken at both the widest and narrowest axes and the diameter was calculated as the square root of the product of the two measurements (Gerwing et al. 2006).

2.6.2 The reference rainforest survey

2 The one-hectare reference plot was subdivided into one hundred 100 mP P grid squares and all tree stems greater than 5cm in DBH were mapped and measured in 1995 (Laidlaw et al. 2000). As part of the present study, a resurvey was carried out in 2006. All remaining stems from the initial survey were re-measured and new stems that had grown into the sampling size classes were also mapped and measured. In addition to the canopy trees (DBH ≥ 5 cm), small trees (1 cm ≤ DBH < 5 cm) and seedlings (Height < 2.5 m) were also surveyed in 2006 using subplots on 50 (odd numbered) of the 100 grid squares prior to the main resurvey. Within each of the chosen grid squares, a 5×5 m subplot was established and the small trees (1 cm ≤ DBH< 5 cm) within the subplot were recorded and identified. Meanwhile, five 1×1-m seedling plots were also marked out with one at each of the four corners and one at the center of the 10 m×10 m plot.

A liana survey in the 1 ha plot was based on the one hundred 10 m×10 m grid squares. This was the first occasion on which the liana had been surveyed on this plot. All the liana stems > 2 cm were identified and recorded using the same methods as for the edge liana surveys. The root and crown positions of each liana stem were also located using X and Y coordinates within the 1ha plot. In addition, the tree(s) climbed by each of the lianas was/ were also recorded for analysis liana and tree associations.

10×10m quadrate

1×1m seedling subplots

40m 20m 20m 40m 60m 100m

Non-rainforest matrix Rainforest (Eucalypt forest or pasture)

Edge

Figure 2.4 Layout of vegetation sampling plots along an edge transect. (Non-rainforest matrices are eucalypt forests and pastures. Sampling efforts for each edge types are presented in table 2.1) 25

2.7 Soil Seed Bank Sampling

This soil seed bank study compared the species composition and seed density of soil seed banks in the top 5 cm of soil. Soil samples were collected using a 10×10×5-cm deep steel corer with sharpened edges at one end. One hundred soil samples were collected from the rainforest reference plot, one sample from the center of each of the 100 grid squares. For the rainforest edges, two soil samples were collected 10 m from each other perpendicular to the edge transect at each of the five distances into rainforest and three distances into pasture and eucalypt forest, repectively. In total, 370 soil samples were collected. Due to the large amount of seedling identification and limited germination space, the soil seed bank sampling and subsequent germination were done as two experiments in two separate years. The reference th th rainforest and the roadside edges (190 samples) were sampled on the 5P andP 8P P of September 2004 respectively and the germination experiment was running over the next 6 months, finished in March 2005. The rainforest/ eucalyptus forest and th rainforest/ pasture edges were sampled on the 15P P of August 2005 and the germination experiment carried out from then until February 2006.

2.8 Germination Experiment

The germination experiment was carried out in a shade house located on the roof of a three storey building at the Nathan Campus of Griffith University. The bush house consisted of a steel frame covered by a shade cloth, allowing about 30 percent penetration of direct sunlight (Measured by EMTEK, EMT-201 Lux meter). An automatic sprinkler system was used in the shade house to water the soil samples.

Soil samples were spread to a depth of about 1 cm over a layer of vermiculite in plastic germination trays. The germination trays were then placed on germination benches in the shade house. Sticky Tanglefoot® was applied to the legs of those germination benches to exclude insects.

Germination observations were conducted twice a week during the experiment and seedlings were removed following identification. Specimens of each species were photographed to assist seedling identification and then were transplanted to obtain

larger specimens for final identification. Large seedlings that were unable to be identified were transplanted to individual pots for identification after further growth. During the experiment, the soils were mixed from time to time to allow buried seed to come to the surface. The germination experiments lasted about 6 months until there was no germination recorded over at least two consecutive weeks.

Five germination trays containing gardening mixture were used to monitor possible seed contamination. Towards the end of the germination experiment, two species, Pilea microphylla and Eucalyptus maculata, were found in the control germination trays. The seeds of E. maculata were sourced from a nearby fruiting tree. Pilea microphylla was found growing at some damp areas around the building. Both species were excluded from analysis.

2.9 Data Analysis

Statistical analyses applied in each chapter have been described separately. One of the primary objects of this study was to identify the patterns with which rainforest plants respond to edge effects, using both multivariate and univariate approaches. For a study of this kind, screening a long list of species using large tables of statistical tests may raise the concern of potential Type I errors - that is: rejecting the null hypothesis more frequently than we should (Rice 1989). The conventional approach to minimize this effect is to lower the alpha level of significance thus reducing the likelihood of inappropriate rejection of null hypotheses, for example, using a sequential Bonferroni correction (Rice 1989). The present study, however, has not applied a Bonferroni correction and has maintained the critical significance level constant of 0.05 to avoid the potential for Type II error: the risk of excluding ecological meaningful results (Moran 2003; Garcia 2004). The same approach has been used in many similar ecological studies (Schowalter et al. 2005; Smart et al. 2005; Baker et al. 2007; Naaf & Wulf 2007). Where possible, the p-values of the statistical tests have been included in the Tables should further judgment be required.

3 RAINFOREST FRAGMENTATION AND TREES

3.1 Introduction

Fragmentation, and the corresponding increase in rainforest edges, have become pervasive phenomena threatening the survival of the remaining rainforest (Whitmore 1997; Laurance & Peres 2006). In addition to the loss of large areas of wildlife habitat, one of the major consequences following rainforest fragmentation is the increase in the extend of edges, exposing rainforest interiors to other, mostly ‘hostile’ influences from surrounding matrices (Lovejoy et al. 1984; Saunders et al. 1991; Murcia 1995). Subsequent changes in the composition and structure of tree assemblages along edges may lead to changes in many edge-related ecological processes which, in turn, can have long term impacts on the regeneration of rainforest near the edges (Murcia 1995; Laurance et al. 1998a; Gascon et al. 2000; Laurance et al. 2000; Laurance et al. 2006b; Schedlbauer et al. 2007).

One of the most significant changes in fragmented rainforest is the dramatic increase in the death of trees especially that of large sized canopy trees, immediately after edge creation. This has been ascribed mainly to the sudden changes in microclimatic conditions, such as an increase in light regimes and temperature, changes in soil and atmospheric moisture, and an increase in the intensity and frequency of wind turbulence (Ferreira & Laurance 1997; Laurance et al. 1998a; Laurance et al. 2000). The forest canopy will become more open near the edges and lead to cascading changes in other rainforest components that rely on the microclimatic conditions and resources provided by the intact canopies, for example the seedling bank, soil seed bank and rainforest animals (Williams-Linera 1990a; Gascon et al. 1999; Laurance et al. 2002; Harper et al. 2005). These dramatic changes during the early stages of edge exposure will bring the rainforest into a highly dynamic state, with increased gap events, accelerated turnover rates, and proliferation of earlier successional species (Laurance et al. 1997; Gascon et al. 2000; Laurance et al. 2006a; Laurance et al. 2006b). The large amount of biomass loss due to the accelerated forest dynamics at edges may even have contributed substantially to the increase in atmospheric carbon dioxide in recent decades (Laurance et al. 1997; Laurance et al. 1998a; Laurance et al.

1998c). Compared with more than three thousand years of human impacts in Europe, Australia has passed from hunter-gathering to modern industrial society in just over 200 years, with concomitant intensification of land use. This short period of intensive human utilization makes the remaining vegetation extremely vulnerable as there has not been enough time for adaptations to these novel forms of disturbances to evolve in the flora (Hobbs & Hopkins 1990). One of the most striking features of the remaining rainforests in Australia is the abrupt boundaries between rainforest and other vegetation types, especially the eucalypt forest (Webb & Tracey 1981; Unwin 1989; Turton & Sexton 1996; Bowman 2000). The occurrence of eucalypt forest and their associated fire regimes hampers the natural regeneration of rainforest and frequently pushes the boundary deeper into the rainforest (Unwin 1989; Floyd 1990; Bowman 2000). Moreover, since European settlement, some of these edges have been exposed to continuing human disturbances, such as grazing, selective clearing and burning, which may well lead to further degradation of the remaining rainforest (Floyd 1990; Laurance 1991a; Kershaw 1992). Despite their importance in rainforest conservation, rainforest edges are understudied and many of the ecological processes associated with rainforest fragmentation are still poorly understood (Laurance & Peres 2006).

In this chapter I compare the composition and structure of the tree flora in a one- hectare rainforest reference plot with that of rainforests close to three types of rainforest edges. The main questions addressed in this chapter are: 1. are there changes in the species composition and structure of the rainforest at edges and are these changes different among different types of edges?; 2. how deeply do edge effects penetrate into rainforest in terms of changes in species composition and structure of tree assemblages?; and, 3. how are different tree species distributed along edges in terms of different regeneration guilds (shade tolerance) and do these patterns vary in response to different types of surrounding matrices?

3.2 Methods

3.2.1 Tree sampling

The one-hectare reference rainforest was subdivided into one hundred 10 m×10 m grid squares (for site description see Chapter 2). Within each of the grid squares, all tree stems above 5 cm DBH were identified and measured. In order to add information about smaller trees, less than 5 cm DBH, a 5 m × 5 m subplot was established in half of the 100 grid squares (odd numbered). In each of these 50 5 m × 5 m subplots, all tree stems bigger than 1 cm and smaller than 5 cm were identified and measured.

The tree survey at rainforest edges was based on the edge transects which stretched 100 m from the edge into rainforest interiors. For each type of edge (eucalypt forest/ rainforest edge, pasture/ rainforest edge and roadside rainforest edge), nine transects were established within the subtropical rainforest complex at Lamington National Park (for edge locations see Chapter 2, Fig. 2.1). Along each transect, five 10 m × 10 m plots were established at intervals (between plot centres) of around 20 m from edge into the rainforest interior. Within each of the sampling plots, all trees above 1 cm DBH were identified and measured.

3.2.2 Data analysis

Tree species diversity indices were calculated using the Shannon diversity index (Magurran 2004), based on trees with DBH > 5 cm. Tree species richness of this forest was estimated by a first-order jackknife estimate using EstimateS (Colwell 2005).

The distributions of the densities and total basal areas of trees were compared using two-way ANOVA, considering edge type and the distance from edge to rainforest interior as independent factors. For these analyses, trees were grouped into four size classes (DBH ≤ 5 cm, 5 cm < DBH ≤ 10 cm, 10 < DBH ≤ 20 cm and 20 cm < DBH) The analysis was carried out for each size class separately since other studies have found that trees in different size classes may respond differently to edge effects (Laurance et al. 1998a; Schedlbauer et al. 2007).

The species composition of tree assemblages in different size classes was compared within each edge type and the rainforest reference plot using Sorensen Similarity Indices (Magurran 2004). Trees were classified as shade-tolerant, intermediate and shade-intolerant, according to their ‘regeneration guilds’ reported in other studies in the same area (Hopkins 1975; Shugart et al. 1980; Smith et al. 2005) and the Flora of New South Wales (Harden 1991). The distributions of the densities and total basal area of trees in different regeneration guilds were compared among different types of edges and different positions from edge to rainforest interiors, using two-way ANOVAs. Whenever a significant difference was identified among edges, a post hoc test using Tukey’s HSD was used to determine which comparisons produced the significant differences.

Comparisons of tree density and basal area were also conducted between the rainforest reference plot and the rainforest edges using one-way ANOVA. Whenever a significant result occurred, a pos hoc test using Tukey’s HSD was applied for pair- wised comparisons. To match the sample size at the edges, 45 plots were randomly chosen from the pool of 100 10 m × 10 m square plots from the rainforest reference plot.

The distribution of tree densities and basal area along the transects from edges to the rainforest interior was further assessed using Spearman Rank Correlations (Quinn & Keough 2002). This analysis was used to assess the distribution patterns of trees at each type of edge, especially when significant interactions between edge type and distance effects were detected by ANOVA.

All of the above statistical tests were carried out on SPSS 13.0 (SPSS Inc. 2004), using logarithmically transformed data for a normal distribution.

Many studies have found that shade-intolerant species are good indicators of disturbance and subsequent changes in microclimatic conditions in rainforest (Denslow 1987; Williams-Linera 1990a; Molino & Sabatier 2001; Laurance et al. 2006b). For this reason, the composition of this particular group of trees was further compared among the edges. A non-metric multi-scaling ordination (NMDS) was

carried out, based on Bray-Curtis similarity matrices calculated among edges using PRIMER 5 (Clarke & Warwick 2001). An analysis of similarity (ANOSIM: Clarke 1993), with 999 permutations, was performed to test the significance of the pattern identified. Edge type and distances from edge to forest interior were considered as fixed factors in the analysis.

The associations between the abundances of each species and ordination coordinates was assessed using multiple regression (Using Pop Tools Excel add-in, Hood 2004), and their significance levels were tested using a randomization test (Edgington 1980). Biplot vectors (Jongman et al. 1995) were then generated for those species with a significant multiple regression result (P<0.05), showing the trends in increasing abundance for those species in the ordination space. Coordinates of each terminal point for those species were the Pearson’s correlation indices between the species abundance and the x and y ordination scores.

3.3 Results

3.3.1 Rainforest in the one-hectare plot

A total of 1334 stems with DBH above 5 cm was recorded on the 1 ha rainforest reference plot, of which 636 stems were bigger than 10 cm. Eighty-five species of trees (DBH >5cm) were identified in this single hectare, of which, there were 66 species with DBH above 10 cm. Numerically, Actephila lindleyi, Atractocarpus benthamianus, Baloghia inophylla, Argyrodendron trifoliolatum and Caldcluvia paniculosa were the most dominant species, which together composed 43.66% of the total tree stems.

2 The total basal area for trees with DBH above 5 cm in this hectare reached 76.2 mP ,P with more than 55% contributed by the five most dominant species, A. trifoliolatum, Ficus watkinsiana, Argyrodendron actinophyllum, C. paniculosa and Pseudoweinmannia lachnocarpa. A. trofoliolatum alone composed nearly 18% of the total basal area.

The additional survey of small trees (5 cm > DBH ≥ 1 cm) contributed only an

additional six species to the total tree species richness, all understorey species, such as Pittosporum multiflorum, Harpullia alata, Triunia youngiana and Linospadix monostachya.

The estimation of species richness, based on trees above 5 cm, suggests that the species encountered on the one hectare plot are drawn from a local pool of more than 106 species. Tree species diversity (H’) was calculated as 3.42 and 3.26 based upon trees with DBH above 5cm and 10 cm respectively, which is similar to that recorded in a nearby part of the rainforest (About 500 meters away, Connell et al. 1984).

In terms of different regeneration guilds, shade-tolerant species dominate this forest, composing nearly 75% of the total number of trees and 73% of total tree basal area, respectively. The shade-intolerant species are also frequently found, making up 9% of both the total tree stems and tree basal area.

Table 3.1 Density, species richness, and diversity of trees within the rainforest reference plot and the rainforest edges. (Densities are pooled averages (± SE) of the tree density at difference distances from rainforest edges) Number of trees / ha Species richness Shannon’s H DBH>1cm DBH>10cm DBH>1cm DBH>10cm (DBH>5cm) Rainforest reference plot 5174 ± 352 636 ± 23 91 66 3.42 Eucalypt forest/ rainforest 4040 ± 191 847 ± 43 126 66 3.92 Pasture/ rainforest 4635 ± 207 1097 ± 61 122 83 3.91 Roadside edge 4447 ± 166 711 ± 34 119 65 3.65

3.3.2 Change of tree assemblages at rainforest edges

A total of 168 tree species, including a few non-rainforest species, was identified at the three types of edges. A similar number of tree species was recorded at each type of edge (Table 3.1). Many species were recorded only as small trees (DBH<10 cm, Table 3.1). The pasture/ rainforest edges had more species as trees above10 cm in DBH than the other two types of edges, which may be explained partly by the higher density - about 20% higher in terms of both tree stems and species - than other two types of edges.

Values of the Shannon-Wiener index were higher at Eucalypt forest/ rainforest edges

and pasture/ rainforest edges than within the 1 ha rainforest or the roadside edges (Table 3.1). The species estimation curves at the edges showed similar patterns, with none of them reached an asymptote (Fig. 3.1).

150 ecie p 100

50 Roadside edge Pasture rainforest edge Eucalypt rainforest edge Reference rainforest

Cumulative number of s 0 0 102030405060708090100 Number of plots

Figure 3.1 Species discovery curves based on first-order Jack-knife estimations of the tree species richness (DBH >5cm) within the rainforest reference plot and the rainforest edges.

3.3.3 Distribution of tree density

The density of all trees showed a significant difference among edges with a strong distance effect (Table 3. 2). An interaction between the two factors, however, suggests that the distribution patterns were different among edges (Fig. 3.2).

The density of saplings (DBH<5 cm) showed a similar pattern to the total trees as this group made up around 60% of all trees recorded (Table 3.2).

The densities of the two groups of small-sized trees (5 cm ≤ DBH< 10 and 10 ≤ DBH < 20 cm) were significantly different among edges and showed a significant distance effect with no interaction between the two factors (Table 3.2). The densities of both groups were higher at the pasture/ rainforest edge than at the roadside edges and neither of these was significantly different from that at the eucalypt/ rainforest edge (Two-way ANOVA, P < 0.01, Tukey’s HSD).

Densities of large-sized trees (DBH ≥ 20cm) were different among edge types but

showed no significant distance effect (Table 3.2). There was an interaction, however, between edge type and distance effect indicating that the distributions may be different at different edges.

4500 8000 Pasture/rainforest edge 4000 7000 Roadside edge 3500 6000 Eucalypt forest/rainforest edge 3000 5000 2500 4000 2000 3000 1500 2000 1000 1000 500 0 0 Reference 0 20406080100 020406080100Reference

1600 All trees 1000 DBH<5cm 900 1400 800 1200 700 1000 600 800 500 400 600 300 400 200 200 100 Number of stems /ha 0 0 Reference 0 20406080100Reference 0 20406080100 5cm≤DBH<10cm 800 700 10cm≤DBH<20cm 600 500 400 300 200 100 0 0 20406080100Reference

20cm≤DBH

Distance from edge to rainforest interiors (m)

Figure 3.2 Changes in the densities of trees from edges to rainforest interiors(average of 9 plots with error bars indicating standard errors) , with comparisons with the rainforest reference plot (average of 100 plots with error bars indicating standard errors).

Spearman Rank Correlations showed that the tree density decreased steadily from edge to rainforest interiors at the pasture/ rainforest edge regardless of the different size classes (Table 3.3). In contrast, the density of trees showed no correlation with the distances for all class at the roadside edges. At eucalypt forest/ rainforest edges, the density of all trees was significantly correlated with the distance but only the saplings (1cm≤ DBH<5cm) followed this general pattern when the trees were grouped

into different size classes.

Compared with the rainforest reference plot, the eucalypt forest/ rainforest edges showed a significantly lower density of all trees (DBH> 1cm, one-way ANOVA, P = 0.003, Tukey’s HSD); both the eucalypt forest/ rainforest and pasture/ rainforest edges showed a lower density of saplings (DBH 1-5 cm, one-way ANOVA, P < 0.001, Tukey’s HSD); and the pasture/ rainforest edges showed a higher density of trees in the two large size classes (DBH 10-20 cm and 20 cm and above, one-way ANOVA, P < 0.001, Tukey’s HSD). Other than those listed above, there are no significant differences between the rainforest reference plot and the edges in terms of tree densities in different size classes.

3.3.4 Distribution of tree basal area

The total tree basal area showed a weak difference between edge types without a significant distance effect or an interaction between the factors (Table 3.2). Post-hoc tests, however, showed no significant difference among edges when they were compared in a pair-wise manner.

Among different size classes, only the total basal area of small trees (5 cm≤ DBH < 10 cm) showed a strong among edge difference (Table 3.2). Within this class, the pasture/ rainforest edges maintained a higher tree basal area than that of eucalypt forest/ rainforest and roadside edge (Fig. 3.3, ANOVA, P< 0.01, Tukey’s HSD).

There was no significant difference detected in terms of either the overall of total tree basal area or the tree basal area of difference tree size classes between the rainforest and the edges (one-way ANOVA, P>0.05).

In terms of total basal area, only the saplings (1≤DBH<5 cm) showed a significant difference among different distances from the edge (Table 3.2). The significant interaction between the edge type and distance effects, however, suggests that the distribution patterns of the total basal area of saplings were different among edges. Apparently, the eucalypt/ forest edges and the pasture/ rainforest edges showed different patterns from the roadside edges (Fig. 3.3).

Table 3.2 ANOVA results of the comparisons of the densities of trees in different size classes at different positions from edges to the rainforest interiors and among different types of rainforest edges (significant results are highlighted). All stem F-ratio P-value 1-5 cm 5-10 cm 10-20 cm 20 cm and above Source Df F-ratio P-value F-ratio P-value F-ratio P-value F-ratio P-value Tree density Edge 2 3.699 0.028 6.204 0.003 9.460 <0.001 6.446 0.002 13.282 <0.001 Distance 4 7.425 <0.001 4.683 0.002 3.977 0.005 4.938 0.001 1.205 0.312 edge * Distance 8 2.157 0.036 2.948 0.005 0.614 0.765 1.651 0.117 2.022 0.049 Basal area

Edge 2 3.432 0.036 0.666 0.516 5.627 0.005 0.191 0.827 1.648 0.197 Distance 4 0.745 0.563 6.583 <0.001 2.248 0.068 1.559 0.190 0.656 0.624 edge * Distance 8 1.487 0.169 2.182 0.033 1.205 0.302 0.783 0.618 1.205 0.302

Spearman Rank Correlations revealed a significant decrease of total basal area of small size trees from the edge to the rainforest interior at pasture/ rainforest edges while that of the large size trees showed an opposite pattern, increasing from edge to rainforest interior (Table 3.3). At eucalypt forest/ rainforest edges, only the total basal area of small saplings (1cm ≤ DBH< 5 cm) showed a significant distance effect, decreasing from the edge towards the rainforest interior. The tree basal area, for all size classes, showed no correlation with distance at the roadside edges.

Table 3.3 Correlation coefficients (Spearman Rank Correlations) between the densities and total basal area of trees with the distances from edge to the rainforest interiors (* P<0.05, ** P<0.01) Tree category Pasture/ rainforest Eucalypt forest/ Roadside/ rainforest rainforest Number of stems All individuals -0.739** -0.436** -0.033 1-5 cm -0.607** -0.409** 0.008 5-10 cm -0.510** -0.210 -0.188 10-20 cm -0.474** -0.034 -0.104 20 cm and above -0.310* 0.276 -0.016 Total basal area All individuals 0.288 0.246 -0.175 1-5 cm -0.636** -0.398** 0.002 5-10 cm -0.445** -0.114 -0.178 10-20 cm -0.440** 0.027 -0.065 20 cm and above 0.385** 0.248 -0.151

3.3.5 Similarity among different size classes

Pasture/ rainforest edges showed a higher species similarity across trees of different size classes than other edges and the rainforest reference plot (Table 3.4). Eucalypt/ rainforest edges showed very low similarity between trees of small size classes (1 cm - 5 cm and 5 cm-10 cm) and large trees (20 cm and above). A similar pattern was found at roadside edges.

Table 3.4 Sorensen Similarity Indices between tree assemblages in different size classes (tree size class: 1. 1 cm - 5 cm; 2. 5-10 cm; 3. 10 - 20 cm and 4. 20 cm and above) 1 vs 2 1vs3 1vs4 2vs3 2vs4 3vs4 Eucalypt/ rainforest edge 69.9 58.3 41.1 67.8 49.5 53.3 Pasture/ rainforest edge 76.7 69.0 58.0 73.6 60.6 66.7 Roadside edge 72.9 55.0 50.7 61.5 50.9 60.2 Rainforest reference plot 68.3 67.9 51.9 69.6 52.7 60.2

140 3.5

120 3

100 2.5

80 2

60 1.5 Pasture/rainforest edge 40 1 Roadside edge 20 Eucalypt forest/rainforest edge 0.5 0 0 020406080100Reference 0 20406080100Reference 7 All trees 18 DBH<5cm 16

a 14 5 12 4 10 8 3 6 2 4 1 2 0 0 Reference 0 20406080100Reference 020406080100 Number of stems / h 120 5cm≤DBH<10cm 10cm≤DBH<20cm 100

0 0 20406080100Reference 20cm≤DBH

Distance from edge to rainforest interiors (m)

Figure 3.3 Distributions of the total basal area (average of 9 plots with error bars indicating standard errors) of trees from the edges to rainforest interiors, with comparisons with the rainforest reference plot (Average of 100 plots with error bar indicating standard errors).

3.3.6 Distributions of trees in different regeneration guilds

The density of shade-tolerant trees differed significant among the edge types without a distinctive distance effect (Table 3.5). Post hoc tests on the edge type suggest that the density of shade-tolerant trees was higher at roadside and pasture/ rainforest edges than at eucalypt forest/ rainforest edges.

Similarly, the density of intermediate species showed no difference among edges and was comparable at different distances from edge to rainforest interiors, although the 1 cm-10 cm size class at pasture/ rainforest was slightly higher (Table 3.5, Fig. 3.4).

6000 Eucalypt forest/rainforest edge 5000 Pasture/rainforest edge Roadside edge 4000

3000

2000 Shade tolerant species 1000

0 Reference 020406080100 1400

1200

1000

800 Intermediate species 600

400 Number of stems / ha 200

0 020406080100Reference 3500

3000

2500

2000

1500 Shade–intolerant species 1000

500

0 Reference 0 20 40 60 80 100

Distance from edge to rainforest interiors (m)

Figure 3.4 Distributions of the densities of trees in three regeneration guilds from edges to rainforest interiors (average of 9 plots with error bars indicating standard errors), with comparisons with the rainforest reference plot (average of 100 plots with error bar indicating standard errors).

The density of shade-intolerant species was significantly different among edges and showed a strong distance effect, but the interaction between the two factors suggested that the distribution patterns differed among edges (Table 3.5, Fig. 3.4). Compared with the reference rainforest, only the pasture/ rainforest edges showed a higher density of shade-intolerant species while the other two types of edges showed

no difference (One-way ANOVA, P<0.001, Tukey’s HSD). There was no significant difference between the reference rainforest and the edges in terms of the density of trees in the other two regeneration guilds.

Table 3.5 Results of two-way ANOVA for the densites of trees in different regeneration guilds at different distances from edges to rainforest interiors and among different types of rainforest edges. Shade tolerant Intermediate Shade intolerant Source df F-ratio P-value F-ratio P-value F-ratio P-value Tree density Edge 2 11.427 <0.001 1.115 0.331 5.280 0.006 Distance 4 0.500 0.736 1.455 0.220 11.004 <0.001 Edge * Distance 8 1.450 0.183 1.263 0.269 3.020 0.004 Basal area Edge 2 0.258 0.773 0.046 0.955 5.693 0.004 Distance 4 4.624 0.002 0.385 0.819 4.011 0.004 Edge * Distance 8 3.250 0.002 1.113 0.360 0.714 0.679

The total basal area of the shade-tolerant species showed no significant differences among edges but a significant distance effect and an interaction between the two factors were found (Table 3.5, Fig. 3.6). The total basal area of intermediate species showed no difference among edges and also showed no distance effects (Table 3.5). The total basal area of shade-intolerant trees was different among edges with a significant distance effect and an interaction of the two factors (Table 3. 5, Fig. 3.6).

Compared with the rainforest reference plot, only pasture/ rainforest edges showed a higher total basal area of shade-intolerant species (One-way, ANOVA, P <0.001, Tukey’s HSD). There was no significant difference between the reference rainforest and the edges in terms of the basal area of trees in both shade-tolerant and intermediate guilds.

Spearman Rank Correlations showed a significant decrease in the density of trees in all the three regeneration guilds from the edge to rainforest interiors at pasture/ rainforest edges (Table 3.6). The total basal area of trees in different regeneration guilds, however, showed different patterns, with that of the shade-tolerant species increasing, the shade-intolerant species decreasing and the intermediate species showing no correlation with the distance from the edge to the rainforest interior.

120 Eucalypt forest/rainforest edge Pasture/rainforest edge 100 Roadside edge

40 Shade tolerant species 20

0 Reference 0 20406080100 30

25 Intermediate species 20

10 Number of stems / ha 5

0 0 20 40 60 80 100 Reference 70

50 40 Shade–intolerant species 30

0 0 20 40 60 80 100 Reference Distance from edge to rainforest

Figure 3.5 Distributions of the basal areas of trees in three regeneration guilds from edges to rainforest interiors (average of 9 plots with error bars indicating standard errors), with comparisons with the rainforest reference plot (Average of 100 plots with error bar indicating standard errors).

Both the density and total basal area of shade-intolerant species showed negative correlations with the distances from edge to rainforest interior whereas the total basal area of shade-tolerant exhibited a positive correlation with distance from the edge at eucalypt forest/ rainforest edges (Table 3.6).

The distributions of all tree, without consideration of their regeneration guilds, showed no significant correlations with the distances from the edge to the rainforest interior (Table 3.6)

Table 3.6 Spearman Rank Correlations between densities and total basal areas of trees in different regeneration guilds with the distances from edges to rainforest interiors. (* P<0.05, ** P<0.01) Regeneration guild Pasture/ Eucalypt forest/ Roadside/ rainforest rainforest rainforest Number of stems Shade-tolerant -0.529** 0.073 -0.021 Intermediate -0.408** -0.138 0.089 Shade-intolerant -0.575** -0.400** -0.175 Total basal area Shade-tolerant 0.540** 0.352* -0.100 Intermediate -0.021 0.104 0.053 Shade-intolerant -0.469** -0.409** -0.188

3.3.7 Composition of shade-intolerant species

The NMDS ordination showed good separation across edge types and this difference in the composition of shade-intolerant species between edge types was statistically significant (ANOSIM Global R = 0.414, P = 0.001). Pair-wise tests also showed that each of the three edge types were different from one another (Table 3.7). Eight species were found to be significantly associated with the two ordination axes; Croton verreauxii, Araucaria cunninghamii and Euroschinus falcatus var. falcatus were associated with eucalypt/ rainforest edges; Acronychia suberosa, Clerodendrum floribundum and Rhodomyrtus psidioides with pasture/ rainforest edges; and Orites excelsus with roadside edges (Fig. 3.6 ).

Table 3.7 ANOSIM results for the comparisons of shade-intolerant species among different types of rainforest edges (Global R = 0.414, P=0.001) Source R Statistic P Eucalypt forest/ rainforest edge vs Pasture edge 0.492 0.003 Eucalypt forest/ rainforest edge vs Roadside edge 0.458 0.003 Pasture edge vs roadside edge 0.31 0.010

3.4 Discussion

Compared with other subtropical rainforest studies (Smith et al. 2005), the rainforest reference plot was extraordinarily diverse in terms of tree species richness, with 86 species of trees above 5 cm in the single hectare studied. The region is located in an area of botanical overlap between the sub-tropical and temperate zone (The

'McPherson-Macleay' botanical overlap, Burbidge 1960) and this may have contributed to the high tree diversity (Laidlaw et al. 2000). The area is also known as one of the centres of high endemicity for Australian biodiversity (Crisp et al. 2001). The total basal area of this plot is much higher than that from other studies of similar rainforest (Burgess et al. 1975; Connell et al. 1984; Smith et al. 2005), which indicates that this forest is in a relatively undisturbed old growth state.

Eucalyptus forest/rainforest Eucalypt forest/rainforest edges Stress=0.15 Pasture/rainforestPasture/rainforest edges RoadsideRoadside/rainforest edges

CRVE ARCU

EUFA ACSU RHPS ACPU CLFL

OREX

Figure 3.6 NMDS analysis based on the composition of shade-intolerant tree species at different edge types. Vectors show species significantly associated with the ordination coordinates. ACPU: Acronychia pubescens; ACSU: Acronychia suberosa; ARCU: Araucaria cunninghamii; CLFL: Clerodendrum floribundum; CRVE: Croton verreauxii; RHPS: Rhodomyrtus psidioides; EUFA: Euroschinus falcatus var. falcatus; OREC: Orites excelsus.

The results also show that the tree assemblages have been largely modified at rainforest edges, compared with rainforest interiors and the rainforest reference plot. A clear distance effect from edge to rainforest interior is regulating the distribution of trees. The responses to this distance effect varied substantially among different species and across the different surrounding matrices at the edges. Significant increases in the abundance of shade-intolerant trees close to the edges were consistent across edge types, but different edges were associated with presented different species

compositions.

3.4.1 Matrix regulated edge effects

In responding to different matrices, rainforest edges generally vary considerably in species composition and structure (Laurance et al. 1998a; Williams-Linera et al. 1998; Mesquita et al. 1999; Denyer et al. 2006; Laurance et al. 2006b). Our results show that the matrix types play important roles in regulating the composition and dynamics of the edges. It was obvious that the pasture/ rainforest edges experienced the strongest edge effects, as they were exposed directly to open light. The high density of medium sized trees showed that vigorous regeneration may still be going on. These tree assemblages were characterized by some shade-intolerant species such as Acronychia spp., O. excelsus, Guioa semiglauca and R. psidioides, which are generally found in the later stages of rainforest succession (Webb et al. 1972; Hopkins 1975; Williams et al. 1993). Since all those pasture/ rainforest edges were well fenced, it is likely that these edges may regenerate towards the old growth rainforest.

The remarkably abrupt Eucalypt/ rainforest edges are recognized as one of the prominent features of Australian rainforest and have attracted much conservation concern (Ash 1988; Unwin 1989; Bowman 2000). These edges are relatively stable and frequently attributed to fire (Webb & Tracey 1981; Floyd 1990; Kershaw 1992). In eucalypt/ rainforest edges in the study area, frequent fire events are still obvious as evidenced by the recently scorched tree barks. Long-term frequent disturbance at eucalypt/ rainforest edges, mainly from fire, caused a significant increase in small trees, especially those of early successional species, such as C. verreauxii, Polyscias elegans and A. cunninghamii. Some of these species particularly A. cunninghamii, may have been promoted by the germination conditions created by the fires (Floyd 1990). The narrow boundary between rainforest and eucalypt forest differed significantly from the rainforest interiors and has been maintained at a stage similar to early rainforest succession.

The roadside edge appeared to be less affected by edge effects and was relatively close, more or less, to the rainforest interiors in terms of structure and species composition. This may be due to the limited canopy openness, contributed by the

minimal tree removal during road construction and subsequent lateral expansions of the tree canopies at both sides of the road. The regeneration of rainforest edges in general is comparable with gap dynamics taking place within interior rainforests (Laurance et al. 1998a; Laurance et al. 2006a). The changes in forest structure caused by the road construction, however, may have long-term impacts on the dynamics of the rainforests near the edges. For instance, the increase in even-aged large stinging trees may have substantial effects long term dynamics on the rainforests along the edges. The changes in other components, such as lianas and soil seed banks (discussed in chapters 3 and 4) may also have long-term impacts on the rainforest regeneration along the road.

3.4.2 The distance effects

The distance, over which the edge effects impact rainforest dynamics, vary profoundly across different types of edges, depending primarily on the contrast in species composition and structure between the two communities from which are derived the edges concerned (Saunders et al. 1991; Murcia 1995; Harper et al. 2005). In response to edge effects, tree assemblages may show considerable changes in species composition and structure from the edge to the rainforest interior (Laurance 1997; Laurance et al. 1998a; Laurance et al. 2006a; Schedlbauer et al. 2007). In the present study, these changes in tree assemblages were mainly confined to a distance of about 50 m from the rainforest edges.

The distance effect was almost undetectable at the roadside edge, in terms of the changes in tree composition. This may have contributed partly by the careful planning of the road, especially the minimized use of large machinery (Jackson & Stephenson 1986). An increase in total basal area of small trees, however, was observed, possibly due to the release of seedling bank caused by the increased light regimes. This minor change, however, can only be detected a few meters from the edge into the rainforest interiors.

The eucalypt/ rainforest edges showed a sharp change in the species composition and structure of rainforest near the edges. The plots at 0-10 m were significantly different from more interior positions, mainly in their small tree assemblages, especially those

of shade-intolerant species. A lower total tree basal area was also distinctive of the 0- 10 m plots. Beyond this distance, the rainforest showed no significant changes, compared with rainforest interiors. The eucalypt forest may have contributed substantially to the maintenance of the edge preventing the edge effects going any deeper into the rainforest.

The pasture/ rainforest edges appeared to have experienced the strongest edge effects. The increases in the abundance of small trees and shade-intolerant species reached nearly 100 m into the rainforest. Direct exposure to open sunlight may have contributed substantially to this deep penetration of the forest. All the pasture/ rainforest edges are well fenced and no further disturbances, other than the creation of the pasture, have been applied. The disturbance received during the edge creation and subsequent prevalence of strong edge effects, however, may have had impacts deep into the rainforest. It is likely that substantial amount of valuable natural timbers were removed from the rainforest near some of the edges during clearing. For example, rose wood Synoum glandulosum, was collected for making furniture and Vitex lignum- vitae was preferred for fencing posts (O'Reilly 1944). This may also have contributed significantly to the deep penetration of the edge effects.

3.4.3 Shade-intolerant trees at rainforest edges

Shade-intolerant tree species are conspicuous components of rainforest and play important roles in rainforest dynamics, especially in regeneration after disturbances (Brokaw 1987; Uhl et al. 1988; Molino & Sabatier 2001; Schnitzer & Carson 2001). One of the most noticeable changes at rainforest edges is the increase in the abundance of the shade-intolerant species, promoted mainly by the increased light regime and disturbances, as well as by soil perturbation and increased gap events near the edges (Laurance 1997; Oliveira-Filho et al. 1997; Laurance et al. 1998a; Laurance et al. 1998b; Laurance et al. 2002; Laurance et al. 2006b). In the present study, a significant increase in the abundance of shade-intolerant trees was found at all the edges and their distribution followed a linear gradient, decreasing in density and total basal area from the edges to the rainforest interiors. Different types of edges tend to be characterized by slightly different suites of species. In general, pasture/ rainforest and roadside edges maintained similar shade-intolerant tree species as rainforest tree-

fall gaps. In contrast, Eucalypt/ rainforest edges appeared to have many species which have survived frequent fires. Most of these shade-intolerant species are fleshy-fruited and can provide a reliable food source for frugivores which, in return, act as seed dispersers. Some species, such as Alphitonia excelsa and P. elegans, may attract up to 20 species of local frugivorous birds (Church 1997). This belt of shade-intolerant trees may have profound impacts on the regeneration of nearby deforested area as well as of the rainforest interiors in general.

3.4.4 The future of rainforest edges

Rainforests may show considerable resilience to anthropogenic edge effects (Phillips et al. 2006). The results from the present study show that the change of forest structure and composition at rainforest edges were confined mostly to within a distance of 50 metres. Beyond this distance, the forest showed no significant differences from old growth rainforests in terms of density, total basal area and species composition of the tree assemblages. If no further disturbances apply, these edges may well be able to regenerate towards the condition of old-growth rainforest.

Australian is one of the few developed countries in which natural rainforest has been well protected and tremendous efforts have been made to restore rainforest habitats (Erskine 2002; Kanowski et al. 2003). Without the help of natural regeneration, however, the costs of restoration can be extremely high ($15000-25000/ ha) without any promise of long-term success. As a result, such projects are not likely to be put into practice over large areas (Erskine 2002). An alternate approach to restoring natural habitats is to encourage natural regeneration, a much more efficient way of rainforest restoration, both economically and ecologically (Erskine 2002; Lamb et al. 2005). The fragmented rainforest landscape and associated disturbances, however, may generate various ecological barriers to natural regenerations. In Australia, a paucity of large areas of secondary forest may be one of such obstacles impeding the rainforest restoration via natural regeneration. This makes rainforest edges extremely important as they may be the few places where a large amount of secondary rainforest trees are preserved.

4 LIANA AND TREE ASSOCIATIONS IN A FRAGMENTED SUBTROPICAL RAINFOREST

4.1 Introduction

Lianas are a conspicuous component of tropical and subtropical rainforest, and comprise around 25% percent of total woody plant diversity (Putz & Chai 1987; Gentry 1991; Perez-Salicrup et al. 2001; Schnitzer & Bongers 2002).Liana abundance and diversity peak in high rainfall tropical sites such as Amazonian lowland rainforest (Gentry 1991; Perez-Salicrup et al. 2001; Schnitzer 2005). Other factors, including soil type, seasonality of rainfall and disturbance, are also important factors determining the local liana distributions (Laurance et al. 2001b; Putz 2004; Schnitzer 2005; DeWalt et al. 2006). Seasonal drought has been proposed as a mechanistic explanation for high liana abundance in seasonal rainforest, in which lianas can compete successfully with trees by accessing ground water using their deep roots and efficient vascular systems (Schnitzer 2005). This allows the lianas to grow much faster than trees during the dry season and lets them take advantage of the high radiation environment in the forest caused by the low cloud cover and reduced canopy cover at the same time that the growth of trees is limited by water stress (Schnitzer 2005).

Lianas have often been described as ‘structural parasites’ of trees and as having negative effects on their hosts by causing them physical and structural damage and by competing for resources, such as water, nutrients and light (Putz 1984a; Stevens 1987; Laurance et al. 2001b). Other significant negative effects, including enlarging the damage by disturbances and arresting subsequent forest regeneration, may also affect rainforest dynamics (Putz 1984b; Putz & Chai 1987; Schnitzer & Carson 2001; Putz 2004; Schnitzer et al. 2004). As a result, liana removal has often been proposed for rainforest management, especially as part of commercial forestry practice (Putz 1984b; Gerwing 2001; Perez-Salicrup 2001; Gerwing & Uhl 2002; Alvira et al. 2004).

Some studies have suggested that there may be species-specific associations between

lianas and trees (Putz 1984a; Carsten et al. 2002). Indeed, some tree species have developed mechanisms to avoid liana infestation by means such as rapid growth, having smooth bark, shedding branches, leaves or bark (Putz 1984a; Allen et al. 1997; Carsten et al. 2002). The fruit type of trees may also be a factor affecting their susceptibility to lianas, as fleshy-fruited trees may be more likely to attract frugivores which potentially may bring in liana seeds to germinate beneath the trees (Carsten et al. 2002). Other studies, however, suggest that there is no reciprocal associations between trees and lianas (Boom & Mori 1982; Perez-Salicrup et al. 2001; Malizia & Grau 2006). Instead, co-occurrence of some lianas and trees is more likely a result of their similar response to common environmental conditions such as soil type, disturbance and forest successional status (Webb 1958; Balee & Campbell 1990; Gentry 1991; Balfour & Bond 1993; Perez-Salicrup et al. 2001; Malizia & Grau 2006). The auto-correlated or clumped distribution of lianas within rainforests also suggests that lianas may be affected by structural factors related to canopy disturbance (Malizia & Grau 2006). In addition, a large proportion of lianas climb onto their host trees by using other smaller trees or existing lianas, thus the characteristics of their current host tree may not reflect the climbing mechanism of the liana concerned (Putz & Mooney 1991; Perez-Salicrup et al. 2001).

In response to increasing rainforest destruction globally, liana assemblages have been modified at local, regional and even global scales (Putz 1984b; Laurance et al. 2001b; Phillips et al. 2002; Wright et al. 2004). Both the abundance and biomass of lianas increases dramatically after rainforest fragmentation (Laurance et al. 2001b), which may lead to increases in the death rate of trees near rainforest edges. Long-term monitoring of lianas has shown the increasing importance of lianas in rainforest

dynamics, possibly due to increased CO2B B levels favoring the growth of lianas (Phillips et al. 2002; Wright et al. 2004; Phillips et al. 2005). Despite the importance of lianas in rainforest dynamics, few studies have looked at how liana assemblages respond to rainforest fragmentation (but see Laurance et al. 2001b), and the roles of lianas in edge dynamics are still poorly understood. (Laurance et al. 2001b)

In the present study, liana assemblages and their association with trees were compared among three types of rainforest edges and a rainforest reference plot within a subtropical rainforest complex. Some key questions about how lianas respond to edge

effects were addressed, namely. 1) Do lianas of rainforest edges differ from that of the rainforest interior? 2) How do lianas change in species composition in response to different disturbance regimes at rainforest edges? 3) Are there species-specific liana/ tree associations and does this relationship change after rainforest fragmentation?

4.2 Methods

4.2.1 Liana sampling

A survey of canopy liana (DBH≥ 2cm) was conducted within the one-hectare rainforest plot and three types of rainforest edges (for liana sampling methods see Chapter 2).

Within the one-hectare reference plot, all canopy lianas were identified and measured in each of the one hundred grid squares. The root position of each liana stem was located to the nearest half metre in terms of the coordinates on the x and y axes of the 100 × 100 m plot. The tree or trees climbed by each liana stem was/ were also recorded for tree and liana association analysis.

The liana survey at rainforest edges was based on the edge transects which stretched 100 metres from each edge into the rainforest interior (for transect layout see Chapter 2). The five 10 m × 10 m plots along the edge transect used for tree surveys were also used for liana studies. Within each 10×10m plot, all liana stems ≥ 2 cm in DBH were identified and measured at 1.3 m above the ground. The trees climbed by each of the lianas were also identified and measured.

For each type of edge: pasture/ rainforest edge, eucalypt forest/ rainforest edge and roadside edge, a total of 45 plots was surveyed on nine transects across the edges.

4.2.2 Data analysis

Liana density and total basal area was calculated for each of the sampling plots. These

data were compared among the three edge types and the rainforest reference plot as well as among different positions from edge to interior. The species richness of lianas was also estimated based on a first-order jackknife estimate using EstimateS V5 (Colwell 2005)

The distribution of liana density and total basal area from edge to rainforest interior was compared among edges using two-way ANOVA, considering both edge types and distances from edge to rainforest interiors as fixed factors. This analysis was conducted for ‘small’ lianas (DBH < 5cm) and ‘large’ or ‘robust’ lianas (DBH ≥ 5cm) separately as other studies have shown that the distribution patterns of lianas in different size classes may vary from edge to forest interior (Laurance et al. 2001b). In addition, comparisons of liana density and total basal area between the reference rainforest and the three types of edges were also conducted using one-way ANOVA. To match up the sampling size with rainforest edges, 45 plots were randomly selected from the pool of 100 plots at the rainforest reference plot. Whenever a significant difference occurred (P<0.05), a post hoc test using Tukey’s HSD was conducted to compare the differences between groups.

The guild classification of lianas with respect to their climbing mechanism was based upon that of Hegarty (1989) and field observations. Essentially, lianas were classified as: z twiners, which climb their host by twining the trunk or branch of their host tree or by twining other existing liana; z root climbers, which use adventitious roots to climb their host; z tendril climbers, which use leaves or branches modified as tendrils as their climbing tools; or z scramblers, which use hooks or other modified organs to assist climbing.

In this classification, twiners were not subdivided into branch or main stem twiners as other studies have done (Putz 1984b; Dewalt 2000; Laurance et al. 2001b) because a large proportion of lianas in this study used other trees or existing lianas to access their host trees. These lianas were free-hanging from the canopy of the host trees. There were always apparent loops formed at the base of these free hanging lianas,

showing that there had been a small tree or a series of small trees which had died after being used by the liana as a trellis to access the current host

Table 4.1 Classifications of the climbing and seed dispersal guilds of lianas (TW: Twiner; TD: Tendril climber; SC: Scrambler; AD: Adventitious root climber) Liana speices Family Climbing guild Seed dispersal agents Austrosteenisia blackii Fabaceae TW Other Austrosteenisia glabristyla Fabaceae TW Other Caesalpinia subtropica Caesalpiniaceae SC Other Cayratia eurynema Vitaceae TD Frugivore Celastrus australe Celastraceae TW Frugivore Celastrus subspicatus Celastraceae TW Frugivore Cissus antarctica Vitaceae TD Frugivore Cissus hyperglanca Vitaceae TD Frugivore Cissus sterculifolia Vitaceae TD Frugivore Clematis glycinoides Raunculaceae TD Wind Derris involuta Fabaceae TW Other Embelia australiana Myrsinaceae TW Frugivore Hibbertia scandens Dilleniaceae TW Other Jasminum dallachii Oleaceae TW Wind Legnephora moorei Menispermaceae TW Other Maclura cochinchinensis Moreaceae SC Frugivore Marsdenia rostrata Asclepiadaceae TW Wind Melodinus australis Apocynaceae TW Wind Morinda jasminoides Rubiaceae TW Frugivore Palmeria scandens Monimiaceae SC Frugivore Pandorea baileyana Bignoniaceae TW Wind Pandorea jasminoides Bignoniaceae TW Wind Pandorea pandorana Bignoniaceae TW Wind Parsonsia fulva Apocynaceae TW Wind Parsonsia longeipetiolata Apocynaceae TW Wind Parsonsia straminea Apocynaceae TW Wind Parsonsia ventricosa Apocynaceae TW Wind Piper novae-hollandiae Piperaceae AD Frugivore Melodorum leichhardtii Annonaceae TW Frugivore Rubus nebulosus Rosaceae SC Frugivore Rubus moorei Rosaceae SC Frugivore Sarcopetalum harveyanum Menispermaceae TW Frugivore Tetrastigma nitnes Vitaceae TD Frugivore Trophis scandens Moreaceae TW Frugivore

Guilds of lianas based on seed dispersal were classified as frugivore-dispersed, wind- dispersed, or ‘other’ dispersal mechanisms, according to the description of fruit or seed characters in the Flora of New South Wales (Harden 1991) and Rainforest Climbing Plants (Williams & Harden 2000).

Comparison of liana density and the proportion of climbing and dispersal guilds between edge types and the rainforest reference plot were conducted using one-way ANOVA. For the reference rainforest, 45 plots were selected randomly from the 100 plots pool to match up the sample size at the rainforest edges. A post hoc test was applied using Tukey’s HSD to compare the differences among groups once a significant difference was detected.

To test whether or not liana compositions varied between different types of edges and at different positions from the edge to the rainforest interior, non-metric multi-scaling ordinations (NMDS) were conducted based on Bray-Curtis similarity matrices calculated among plots, using the abundances of each liana species. Site type and distance from edge were considered as fixed distribution factors in the analysis. Analysis of similarity (ANOSIM, Clarke & Ainsworth 1993), with 999 permutations, was performed to test the significance of the patterns identified. Twenty-five plots containing no lianas were excluded from the analysis. The test was carried out using PRIMER 5 (Clarke & Warwick 2001).

The associations between the abundances of each liana species and the ordination coordinates were assessed using multiple regression in ‘Pop Tools Excel add-in’ (Hood 2004), and their significance levels were calculated using a randomization test (Edgington 1980). Biplot vectors were added to the coordination plot for those species with a significant (P<0.05) multiple regression result, showing the trends in increasing abundance for those species in the ordination spaces. Coordinates of each terminal point for these biplots were the Pearson’s correlation indexes between the species abundance and the ordination coordinates.

Distribution patterns of lianas among trees were examined by comparing the observed frequency of trees bearing different number of lianas with an expected Poisson distribution (Laurance et al. 2001b). A chi-squared goodness of fit test (Quinn &

Keough 2002) was then applied to test whether or not the liana distribution patterns at the edges and the rainforest interior differed from an expected Poisson distribution.

To test if there was any preference of liana species for climbing particular tree species or for tree species to carry certain species of lianas, a cross table of lianas and infested trees was generated. The recorded number of each liana species on each of the tree species was then compared with a random distribution using a Chi-square goodness of fit test (Quinn & Keough 2002). The same test was also carried out for the stem of each tree species carrying each of the liana species.

To test whether or not study sites with similar tree compositions were also similar in liana compositions, we constructed Bray-Curtis similarity matrixes calculated across study site for trees and lianas separately. A Mantel Test was used to assess the relationship between the similarity matrixes (Quinn & Keough 2002). Given that other studies have showed significant increases in lianas and small trees at rainforest edges (Williams-Linera 1990b; Laurance et al. 1998a; Laurance et al. 2001b), trees were grouped into different size classes and lianas associations with each of the size classes was tested separately.

To evaluate relationships between lianas and the composition and structure of tree assemblages, Spearman Rank Correlations were calculated based on logarithmic transformed densities and total basal area of lianas with the density, species richness and diversity of trees. This correlation was performed for big trees (DBH> 10 cm) and small trees (DBH< 10 cm) separately. The separation is justified on the basis that it may be the small trees and, by inference, their associated lianas which are responding to the edges (Laurance et al. 2001b). This separation was also used to test the hypothesis that the availability of a suitable ‘trellis’ may affect local liana abundance (Putz 1984b).

4.3 Result

4.3.1 Liana composition

In total 32 species of lianas were recorded in this study, including those from the 1 ha

reference rainforest plot and the three types of rainforest edges (Table 4.2). The most abundant liana species were Melodinus australis and Austrosteenisia glabristyla which together accounted for c. 38% of all liana stems.

Liana density was 283 per ha in the 1-ha rainforest reference plot, which was significantly lower than the average of 542 lianas per ha at rainforest edges (Student T-test, t<0.01). Large lianas were not common in any of the sites and their densities ranged from 10 to 24 stems in a hectare. Around 25% to 28% of the trees above 10 cm were climbed by at least one liana across all sites. The highest liana infestation rates were found at the roadside edges (Table 4.2). Compared with the rainforest reference plot, liana density nearly tripled at the pasture/ rainforest edges but almost similar proportions of trees were infested. This indicates that more trees were heavily loaded with multiple lianas at pasture/ rainforest edges.

Some of the lianas were found climbing multiple (up to four) trees, but most of the lianas (90%) were confined within a single tree canopy. The proportion of lianas climbing multiple trees was higher at roadside edges and in the rainforest reference plot (23% and 19% respectively) compared with the pasture/ rainforest and eucalypt forest/ rainforest edge (13% and 9% respectively). On the other hand, a large proportion of liana infested trees were found to bear multiple lianas (up to 13). This was higher at eucalypt forest/ rainforest edges and pasture/ rainforest edges (57% and 55%, respectively) than at either road side edges or the reference rainforest (46% and 42%, respectively).

Table 4.2 Density and total basal area of lianas and the infestation rate of trees at rainforest edges and the rainforest reference plot. (Total areas measured were 1ha for the reference rainforest and 0.45 ha for the edges) Species Density Big liana Total basal Infestation rate Stems/ ha* (>=10cm) area* (Tree>10cm) Reference rainforest 20 283 (26) 10 0.566 (0.07) 0.25 Roadside 27 364 (46) 13 0.766 (0.14) 0.28 Eucalypt forest/ rainforest 19 487 (66) 24 1.138 (0.14) 0.25 Pasture/ rainforest 28 709 (81) 15 1.201 (0.14) 0.27 *Mean (SE)

Roadside and pasture/ rainforest edges had more species of liana than the eucalypt forest and reference rainforest (Table 4.2). Estimates of liana species richness (Fig.

4.1) showed that the eucalypt forest/ rainforest edges were the most species-poor with only 21 species predicted, most of which had been encountered in the first ten plots sampled.

30 ecies p 20 Roadside edge Pasture/rainforest edge 10 Eucalypt forest/rainforest edge Number of s Reference rainforest plot 0 0 20406080100 Number of 10 m ×10 m plot

Figure 4.1 Liana species accumulation curves based on first order jack-knife estimations for the rainforest edges and the rainforest reference plot.

Pasture/ rainforest edges were the most species rich sites with 34 species predicted. The estimation for roadside edges, however, had not reached an asymptote in contrast to the other two sites and more liana species may be expected if sampling effort were to be increased. Both the pasture/ rainforest edges and roadside edges were estimated to have more liana species than the reference rainforest. The curve for eucalyptus forest/ rainforest edges, however, reached an asymptote of 21 species and may be expected to have less species than the rainforest reference plot.

4.3.2 Changes in liana density

The liana densities were significantly different among edge types and showed a strong distance effect without an interaction between the two factors (Table 4.3, Fig. 4.2). Post hoc tests suggested that the pasture/ rainforest edges maintained a higher liana density than did the eucalypt forest/ rainforest and roadside edges, whereas there was no significant difference between the latter two. The liana density was much lower in the reference rainforest than at either eucalypt forest/ rainforest or pasture/ rainforest

edges but was similar to that at the roadside edges ( One-way ANOVA, P<0.001, Tukey HSD).

1400 Roadside edge 1200 Total liana Pasture/rainforest edge 1000 Eucalypt forest/rainforest edge 800

600

400

200

0 Reference 0 20406080100

450 400 350 . Large liana (DBH≥5cm) 300 250 200 150 100 Number of lianas /ha of lianas /ha Number 50 0 0 20406080100Reference

1000 900 800 Small liana (DBH<5cm) 700 600 500 400 300 200 100 0 Reference 0 20406080100 Distance from edge to rainforest

Figure 4.2 Distributions of liana densities from edges to rainforest interiors (average of 9 plots with error bars indicating standard errors), with comparisons with the rainforest reference plot (average of 100 plots with error bar indicating standard errors).

The differences between edges, however, was mainly due to small lianas (DBH<5cm, Fig 4. 2) which showed a significant distance effect and were significantly higher in abundance at the pasture/ rainforest edge than at other two types of edges (Two-way ANOVA, P < 0.001, Tukey’s HSD). The density of these small size lianas in the reference rainforest was comparable to that at the eucalypt forest/ rainforest and

roadside edges but was much lower than at pasture/ rainforest edges (One-way ANOVA, P < 0.001, Tukey’s HSD).

The density of lianas showed significant correlations with the distances from the edge to rainforest interiors at all three types of edge (Table 4.4). These correlations, however, were contributed mainly by small size lianas (DBH < 5 cm) as the large size lianas (DBH ≥ 5 cm) showed no significant correlation with the distances (Table 4.4).

Table 4.3 Results of the two-way ANOVA for the densities and total basal area of lianas at different positions from edges to rainforest interiors and among three types of rainforest edges (significant results are highlighted). All lianas Small lianas (2-5 cm) Large lianas (≥ 5cm) Source Df F-ratio P-value F-ratio P-value F-ratio P-value Density Edge 2 7.744 0.001 8.555 P<0.001 0.616 0.542 Distance 4 6.465 P<0.001 5.580 P<0.001 5.147 0.001

Edge * Distance 8 0.201 0.990 0.277 0.973 0.827 0.580 Basal area Edge 2 2.935 0.057 7.51 0.001 1.20 0.304 Distance 4 4.098 0.004 5.53 P<0.001 2.29 0.064 Edge * Distance 8 0.543 0.822 0.21 0.989 0.71 0.679

4.3.3 Change in liana basal area

Performing the same comparisons as for liana density, but using the total basal area of lianas instead, demonstrated a slightly different distribution pattern (Table 4.3, Fig. 4.3). The total basal area of lianas showed no significant difference among edges but showed a significant distance effect (Table 4.3). The total basal area of lianas at the edges also did not differ significantly from that within the rainforest reference plot (One-way ANOVA, P=0.119).

Despite being similar in total liana basal area, the edges showed a significant difference in the basal area of small lianas (Table 4.3). Post hoc test showed that the pasture/ rainforest edge had a significant higher basal area of lianas than that of the other two types of edges. There was also a significant distance effect. Compared with the reference rainforest, only pasture/ rainforest edges showed a higher total basal area of these small lianas (One-way ANOVA, P < 0.009, Tukey’s HSD).

2.5 Roadside edge

2 Total liana Pasture/rainforest edge Eucalypt forest/rainforest edge 1.5

0.5

0 Reference 0 20 40 60 80 100

1.8

/ ha) / ha) 1.6 2 1.4 . Large liana (DBH≥5cm) 1.2 1 0.8 0.6 0.4

Number of lianas /ha 0.2 Liana basal area (m 0 Reference 0 20 40 60 80 100

1 0.9 0.8 Small liana (DBH<5cm) 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 20 40 60 80 100 Reference Distance from edge to rainforest

Figure 4.3 The distributions of liana basal area from edges to rainforest interior (average of 9 plots with error bars indicating standard errors), with comparisons with the rainforest reference plot (Average of 100 plots with error bar indicating standard errors).

Large lianas (DBH≥5cm) comprised only 20% of total liana stem but 68% of total basal area. The total basal area of large lianas showed no difference among edges or among different positions from the edge to rainforest interiors (Table 4.4). Both eucalypt forest/ rainforest and pasture/ rainforest edges showed a higher total basal area of these large lianas than the rainforest reference plot. The basal area of large lianas, however, was similar at the roadside rainforest edges and the reference rainforest plot (One-way ANOVA, P=0.023, Tukey’s HSD).

Table 4.4 Spearman Rank Correlations of density and total basal area of lianas with the distances from the edges to the rainforest interiors. (* P<0.05, * P<0.01) Eucalypt forest/ Pasture/ rainforest Roadside/ rainforest edge edge rainforest edge Density of liana Total stems -0.378* -0.362* -0.440** 2-5 cm -0.367* -0.330* -0.429** 5 cm and above -0.283 -0.258 -0.231 Basal area of liana Total stems -0.256 -0.274 -0.339* 2-5 cm -0.359* -0.349* -0.411** 5 cm and above -0.168 -0.192 -0.264

The total basal area of lianas showed no correlation with the distances from the edge at eucalypt forest/ rainforest edges and pasture/ rainforest edges, but a significant correlation was found with the distance at the roadside/ rainforest edges (Table 4.4). The small size lianas showed consistent correlations with the distances from the edge to rainforest interior at all three edges (Table 4.4). In contrast, the total basal area of large lianas showed no correlation with the distances from the edge to rainforest interiors.

Table 4.5 ANOSIM results for the compositions of lianas at different positions from edges to rainforest interior and among the rainforest edges. (Significant results are highlighted) Test type Globe R P Edges Eucalypt forest/ rainforest vs Pasture/ rainforest 0.147 0.014 Eucalypt forest/ rainforest vs Roadside edge 0.277 0.001 Pasture/ rainforest vs Roadside ege 0.074 0.070 Distances (m) 0-10 vs 20-30 0.054 0.161 0-10 vs 40-50 0.015 0.341 0-10 vs 60-70 0.155 0.016 0-10 vs 90-100 0.092 0.048 20-30 vs 40-50 -0.012 0.541 20-30 vs 60-70 0.101 0.074 20-30 vs 90-100 0.087 0.096 40-50 vs 60-70 -0.013 0.537 40-50 vs 90-100 -0.006 0.499 60-70 vs 90-100 -0.008 0.535

4.3.4 Comparison of liana composition

MDS results of the liana compositions showed a good separation between roadside edges and eucalypt forest/ rainforest edges (Fig. 4.4, Global R = 0.299, P=0.002).

Further analysis of similarities among edge types showed that the eucalypt forest/ rainforest edges were significantly different from both the roadside rainforest and pasture/ rainforest edges. The latter two did not differ significantly from each other (Table 4.5).

1.5 Stress = 0.16 Roadside/rainforest edge Pasture/rainforest edge Eucalypt forest/rainforest edge 1

0.5

-0.5

-1

-1.5 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

8 9

7 6

5 3 4 2 10 1

Figure 4.4 MDS comparisons of liana composition among rainforest edges, with vectors showing increase in the abundance of liana species significantly associated with the ordination axes. Liana species are:

1. Parsonsia straminea; 2. Melodorum leichhardtii; 3. Maclura cochinchinensis; 4. Caesalpinia subtropica; 5. Tetrastigma nitens; 6. Cissus Antarctica; 7. Pandorea pandorana; 8.Derris involuta; 9. Austrosteenisia glabristyla; 10. Melodinus australis.

A marginal distance effect was detected in the composition of lianas at edges (Global R = 0.027, P=0.048). Further pair-wise tests showed that only those at 0-10 metres were significantly different from plots at 70-80 metres and 90-100 metres; no significant differences were detected among other distances (Table 4.5).

Ten species of lianas showed significant associations with the ordination coordinates (Fig. 4.4). Two species, A. glabristyla and M. australis showed associations with the roadside edges. Eight other species were mainly associated with the eucalypt forest/ rainforest edges and pasture/ rainforest edges.

4.3.5 Comparison of liana climbing guilds

The density of tendril climbers and twiners showed a significant difference among edges whereas the densities of lianas of the other two climbing guilds were similar across the three types of edges (Table 4.6). Only the density of tendril climbers exhibited a significant distance effect and was significantly correlated with the distances from edge to rainforest interior (Spearman Rank Correlation, R=0.340, P<0.001).

Table 4.6 Results of the two-way ANOVA for the densities of lianas in different climbing guilds from edges to the rainforest interiors among different types of rainforest edges. Adventure Scrambler Tendril climber Twiner Source df Root climber F-ratio P-value F-ratio P-value F-ratio P-value F-ratio P-value Edge 2 0.627 0.536 1.696 0.188 4.874 0.009 7.196 0.001

Distance 4 1.928 0.110 1.508 0.204 5.232 0.001 1.812 0.131 edge * Distance 8 0.998 0.449 0.625 0.755 0.361 0.939 0.457 0.884

The densities of twiners were significantly higher at pasture/ rainforest edges than the other two types of edges and the reference rainforest (Fig. 4. 5). The density of scramblers was also higher at pasture/ rainforest edges than at the reference rainforest plot. The densities of tendril climbers were significantly higher at both eucalypt forest/ rainforest and pasture/ rainforest edges than at the roadside edges and the reference rainforest (Fig. 4.5).

600 b 500

400 A a 300 a a a 200 a a 100 b ab ab b b 0 Eucalyptus forest /rainforest

Pasture/rainforest

Roadside rainforest

1 Rainforest reference site c 0.9 0.8 B ab b 0.7 a 0.6 0.5 0.4 a 0.3 b a 0.2 b ab

Proportion Number of lianas/ha ab 0.1 b b 0 Adventure root climber Tendril climber Scrambler Twiner

Figure 4.5 Comparisons of the density (A) and proportion (B) of lianas in four climbing guilds among rainforest edges and the rainforest reference plot.

Twiners made up more than 80 percent of all the lianas in the rainforest reference plot, which was significantly higher than at the rainforest edges (Fig. 4.5). The proportion of twiners was the lowest at the eucalypt forest/ rainforest edges. In contrast, the proportion of tendril climbers showed a steady increase from the rainforest reference plot to the edges and peaked at the eucalypt forest edge (Fig. 4.5). The proportion of scramblers at roadside edges was slight higher than at the reference rainforest but did not differ significantly from other edges.

4.3.6 Comparison of liana dispersal guilds

The density of both frugivore and wind-dispersed lianas were different between edge types, while the density of lianas using other forms of dispersal was similar among the edges (Table 4.7). The densities of frugivore-dispersed lianas were also significantly

different at different positions from the edge and were negatively correlated with the distance from the edge to rainforest interiors (Spearman Rank Correlation, R = 0.348, P<0.001). Densities of other types of lianas showed no significant differences among different positions from the edge to rainforest interiors.

Table 4.7 Results of the two-way ANOVA for the density of lianas in different dispersal guilds at different positions from edges to the rainforest interiors and among different types of rainforest edges (significant results are highlighted). Frugivore-dispersed Other Wind dispersal Source Df F-ratio P-value F-ratio P-value F-ratio P-value Edge 2 7.908 0.001 0.356 0.701 6.231 0.003 Distance 4 5.440 <0.001 2.355 0.058 1.934 0.109 Edge * Distance 8 0.480 0.868 0.983 0.452 0.525 0.836

450 400 a a 350 A 300 b 250 ab ab 200 b a 150 100 b Number of lianas / ha 50 0

Eucalyptus forest /rainforest

0.7 Pasture/rainforest b 0.6 a Roadside rainforest a ab 0.5 Rainforest reference site ab a 0.4 B

0.3 a b 0.2 Proportion 0.1

0 Frugivore dispersed Other dispersed Wind dispersed

Figure 4.6 Comparisons of the density (A) and proportion (B) of lianas in different seed dispersal guilds between rainforest edges and the rainforest reference plot.

The density of wind-dispersed lianas was significantly higher at pasture/ rainforest edges than at the eucalypt forest/ rainforest edges; the density of this guild at roadside edges and the rainforest reference plot was intermediate (Fig. 4.6). The densities of frugivore-dispersed lianas, however, were significantly higher at pasture/ rainforest and eucalypt forest/ rainforest edges than at the roadside edges and the rainforest reference plot. The densities of lianas using other forms of dispersal showed no significant difference among the edges and the reference rainforest.

Wind-dispersed lianas dominated the rainforest reference plot, making up nearly 60% of liana stems. The proportion of these wind-dispersed lianas decreased dramatically from the rainforest reference plot to only 20 percent at eucalypt forest/ rainforest edges (Fig. 4. 6). In contrast, frugivore-dispersed lianas increased significantly from the rainforest reference plot to the rainforest edges, peaking at eucalypt forest/ rainforest edges where more than half of lianas were frugivore-dispersed. Lianas, which were neither frugivore nor wind-dispersed, maintained a similar proportion across all study sites.

4.3.7 Tree and liana associations

Of the 1852 trees >10 cm in DBH recorded in the study plots, only 472 trees were liana infested, giving a total infestation rate of 25.5%. The numbers of lianas borne by each tree stem varied from 1 to 13 with an average of 2.21 lianas per tree within the liana-carrying trees.

Among the 129 tree species recorded in the survey, 83 of them had at least one individual carrying liana. The other 45 liana-free tree species were also low in density with less than five individuals recorded in the survey. The six most dominant tree species, Argyrodendron trifoliolatum, Baloghia inophylla, Pseudoweinmannia lachnocarpa, Diospyros pentamera, Caldcluvia paniculosa and Argyrodendron actinophyllum, altogether comprised 41 % of the overall tree stems and carried 45% of the total lianas.

Compared with a random distribution, tree species showed no significant difference in 2 the probability of being infested by lianas (XP =91.55,P df = 83, p=0.244). In addition, the number of lianas carried by each tree species was significantly correlated with the number of tree stems recorded (R=0.940, P<0.001, Spearman Rank Correlation).

Compared with an expected Poisson distribution, lianas showed a clumped distribution among trees at both the rainforest reference plot and the edges (Fig. 4.7). There were more trees than expected free of lianas, less trees than expected bore small number of lianas (1 or 2), and more trees than expected were climbed by multiple lianas (3-12). This distribution of lianas was not significantly different between the 2 reference plot and the edges (XP =0.21,P df=5, P=0.999).

Table 4.8 The associations of dominant tree species with lianas (trees with> 30 stems or carrying more than 20 lianas in the survey are listed). Total Infested Infestation Number of Liana P* Species stems stems rate liana borne species Argyrodendron trifoliolatum 230 59 0.257 149 21 0.681 Pseudoweinmannia lachnocarpa 97 41 0.423 123 24 <0.001 Baloghia inophylla 158 32 0.203 55 15 0.012 Caldcluvia paniculosa 79 24 0.304 52 11 0.0237 Argyrodendron actinophyllum 82 24 0.293 48 16 <0.001 Polyscias elegans 50 17 0.340 45 14 <0.001 Vitex lignum-vitae 34 17 0.500 42 14 <0.001 Diospyros pentamera 94 19 0.202 35 11 0.157 Atractocarpus benthamianus 71 17 0.239 28 7 0.022 Guioa semiglauca 25 11 0.440 22 8 <0.001 Alphitonia excelsa 16 9 0.409 21 6 0.164 Halfordia kendack 22 7 0.438 21 6 <0.001 Anthocarapa nitidula 25 8 0.32 20 8 0.006 Doryphora sassafras 52 8 0.154 16 4 - Orites excelsus 51 9 0.176 16 7 - Dendrocnide excelsa 32 7 0.219 14 7 - Acronychia suberosa 31 7 0.226 13 8 - Melicope micrococca 32 7 0.219 12 7 - Streblus brunonianus 36 3 0.083 4 7 - * Chi-square test between recorded stem of each liana species on each of the tree species and expected number based on a random distribution.

The infestation rates were different among the tree species from as high as 50% for Premna lignum-vitae to as low as 8.3% for Streblus brunonianus, despite similar numbers of individuals being recorded for these two species (Table 4.8). Most of the tree species showed some preference of carrying certain species of lianas when the recorded number of each liana species was compared with a random distribution (Table 4.8). For example, P. lachnocarpa, the second ranked liana carrier, was more

susceptible to Cissus antarctica and Pandorea pandorana. More than twice as many stems of these two liana species than expected were recorded climbing P. lachnocarpa. On the other hand, there was less than the expected number of M. australis climbing P. lachnocarpa. There were three tree species, A. trifoliolatum, Halfordia kendack and D. pentamera, which showed no significant difference in the likelihood of carrying different species of lianas. The total number of liana species associated with each tree species was also significantly correlated with the total tree stems recorded (R=0.717, P<0.001, Spearman Rank Correlation).

600 800 500 A 700 B 600 400 500 300 400 200 300 Number of trees 200 100 100 0 0 012>2 012>2

Number of lianas/ tree

Figure 4.7 Observed numbers of lianas (bars) carried by rainforest trees within the rainforest reference plot (A) and rainforest edges (B), fitted with expected Poisson distribution values (circles and lines)

Of the 1044 lianas recorded, some were found climbing multiple (up to four) trees (10.3%), but most of them were confined within only a single tree canopy. The number of tree species associated with each of the liana species varied from 1 to 49, which was largely determined by the total number of individuals recorded (R=0.979, P<0.001, Spearman Rank Correlation).

Stems of each tree species associated with each liana species were significantly different from an expected number based on random distributions showing, to some extent, a preference of lianas for climbing particular tree species (Table 4.9). For example, the most dominant species, M. australis, climbed twice as many Atractocarpus benthamianus and H. kendack than expected from a random

distribution but fewer of P. lachnocarpa and A. trifoliolatum than expected. Only Parsonsia straminea showed no preference across tree species.

Table 4.9 Associations of dominant liana species (with >20 stems) with canopy trees. Total Tree species Number of Liana species stems associated tree infested P* Melodinus australis 255 49 183 <0.001 Austrosteenisia glabristyla 143 38 112 0.007 Parsonsia velutina 84 26 66 0.001 Tetrastigma nitens 75 27 55 <0.001 Trophis scandens 73 20 50 0.009 Cissus antarctica 68 22 49 <0.001 Derris involuta 47 20 38 <0.001 Morinda jasminoides 46 18 42 <0.001 Pandorea pandorana 29 11 21 <0.001 Parsonsia straminea 28 13 28 0.141 Celastrus subspicatus 27 15 25 0.001 Cissus sterculifolia 26 16 21 <0.001 Melodorum leichhardtii 23 11 13 0.002 *Chi-square test between recorded stem of each tree species climbed by each of the liana species and expected numbers based on a random distribution.

In the 1 ha rainforest reference plot, significant positive correlations of total basal area of liana and the species richness and total basal area of trees were found (Spearman Rank Correlation, P<0.05, Table 4.10). In addition, the tree diversity was also positively correlated with both the number of stems and the total basal area of lianas.

Table 4.10 Correlations between lianas and trees at rainforest edges and within the rainforest reference plot. (45 plots each edge type and 100 plots at rainforest reference plot, * P<0.05, *P<0.01) Total tree Num. of Tree Num. of Num. of basal area tree species diversity trees > 10cm trees <10 cm Eucalypt forest/ rainforest edge Number of Liana -0.068 0.345* 0.206 0.222 0.310* Liana basal area -0.113 0.528** 0.422** 0.284 0.498** Pasture/ rainforest edge Number of Liana -0.089 0.378** 0.208 -0.038 0.444** Liana basal area -0.206 0.491** 0.358* 0.048 0.518** Roadside/ rainforest edge Number of Liana 0.099 0.262 0.447** 0.172 0.197 Liana basal area 0.024 0.243 0.399** 0.179 0.170 Reference rainforest Number of Liana 0.183 0.192 0.222* 0.123 -0.029 Liana basal area 0.213* 0.228* 0.254* 0.159 -0.102

Among the edges, tree diversity was significantly correlated with the total basal area and number of stems of liana at pasture/ rainforest edges but only correlated with the total basal area of lianas at the other two types of edges. The number of tree species was also correlated with lianas at eucalypt forest/ rainforest and pasture/ rainforest edges but not at the roadside rainforest edges. Similarly, the number of small trees was significantly correlated with lianas at two types of edges except the pasture/ rainforest edges. The number of large trees and the total tree basal area showed no correlation with lianas.

Table 4.11 Results of the Mantel tests for the comparisons of the similarity matrices between assemblages of trees and lianas at rainforest edges (135 plots from all three types of edges). Test type Tree size class (cm) R P Liana vs. all trees DBH≥1 0.65 0.004 Liana vs. small trees 10>DBH≥1 0.75 0.001 Liana vs. big trees DBH≥10 0.85 0.001

A Mantel test, using the pooled data from all three edge types, showed a significant correlation between the similarity matrices of lianas and trees at edges, showing that plots with similar tree composition also had similar liana composition. Similarity matrices of lianas and trees in different size classes were also significantly correlated with each other, regardless of whether the tree similarity matrices were based on all trees, small trees or big trees (Table 4.11). At the reference plot, however, there was no such relationship (Rho=-0.074, p=0.93).

4.4 Discussion

Liana assemblages were largely modified at rainforest edges in comparison with assemblages within the rainforest interiors. This modification was however only detectable within 50 meters of rainforest edges, mainly due to the increases in small lianas. Both liana densities and total basal area were negatively correlated with the distance from the edge to the rainforest interior. Liana abundance and species composition were also varied between different types of edges suggesting that surrounding matrices played important roles in liana dynamics. The contrast in lianas between rainforest interiors and edges also included changes from twiner to tendril climbers and from wind-dispersed to frugivore-dispersed lianas.

4.4.1 Liana distribution at rainforest edges

Many of the changes in environmental conditions from rainforest edge to interior are confined to a distance of 50 meters from the edge, although this distance can vary with different disturbance regimes, disturbance intensity, the surrounding matrix, and the age of the edges (Williams-Linera 1990a; Murcia 1995; Turton & Sexton 1996; Williams-Linera et al. 1998). In response to these changes in environmental conditions, vegetation structure and biotic composition also tend to from edge to forest interior (Williams-Linera 1990b; Laurance 1991b, 1997; Laurance et al. 1998a; Laurance et al. 2001b; Tang et al. 2003). The results of the present study are consistent with these general patterns, with lianas responding to edge effects over a relatively small scale. In particular, changes in liana abundance and total basal area were only detectable within about 50 metres from the rainforest edge. The increase in liana abundance at the edge was mainly a result of an increase in the occurrence of small lianas (DBH< 5 cm) whereas the number of big lianas (DBH≥ 5 cm) showed no difference from the edge to the rainforest interior. The same result was also reported from fragmented Amazon rainforest, in which small liana increased significantly near edges (Laurance et al. 2001b).

4.4.2 Matrix regulated liana communities?

Surrounding matrices have been found to have a strong impact on the dynamics of rainforest edges by regulating abiotic environmental factors as well as edge-related ecological processes (Murcia 1995; Laurance 1997; Williams-Linera et al. 1998; Mesquita et al. 1999). Liana compositions in the present study were significantly different among edge types despite the fact that they shared almost the same regional species pool. Both variation in lateral light penetration and ongoing disturbances may have been responsible for these difference among edge types. For example, the pasture/ rainforest edges, which were directly exposed to a high environment, had the most abundant and diverse liana assemblages at the edges sampled. At this edge type, changes in lianas from the edge to rainforest interior, in both density and total basal area, have penetrated further into the rainforest than at the other two types of edges. This pattern is similar to the changes in small tree assemblages (Chapter3). A similar pattern was found in edge-related tree mortality in Amazonian rainforest fragments, in

which edge effects were found to penetrate further into pasture-bordered edges than edges bordered by re-growth forest (Mesquita et al. 1999).

In terms of changes in species composition, disturbance may play an important role at rainforest edges. Long-term exposure to disturbance from fire and cattle grazing at eucalypt forest/ rainforest edges have maintained a liana community that differed substantially from other two types of edges and the rainforest interior containing, as it did, a large number of early successional liana species. Roadside edges were less exposed and received minimum structural damage, thus the lianas there were similar to the reference rainforest in terms of both species composition and abundance. The increase in the abundance of lianas at the 0-10m sites at roadside edges, however, was significant, which may have been promoted by ramet proliferation of existing lianas and lateral movements of lianas from rainforest interiors. Increases in lianas at forest edges may be comparable with that in forest re-growth as both habitats can promote liana infestation through elevated light and increased availability of small trees as climbing trellises (Putz 1984b; Laurance et al. 2001b). Liana dynamics at edges, especially changes from the edge to rainforest interior, show similar patterns to those found during forest succession, during which liana density decreases from young secondary forest to old growth rainforest while total liana basal area is maintained at a constant level (Vidal et al. 1997; Dewalt 2000).

4.4.3 Liana and tree association

Lianas showed a strongly clumped distribution among trees in the present study with more trees than expected bearing multiple lianas. Similar distributions have been observed in many studies (Putz 1984b; Laurance et al. 2001b; Perez-Salicrup et al. 2001; Schnitzer & Carson 2001; Ibarra-Manriquez & Martinez-Ramos 2002). Two possible mechanisms have been proposed to explain this distribution. First, lianas have been found to be closely associated with disturbance, both natural and anthropogenic (Webb 1958; Gomez-Pompa et al. 1972; Putz 1984b; Putz & Chai 1987; Schnitzer & Carson 2001); many these disturbances are spatially discrete (e.g. treefall gaps). Second, trees that have been infested by lianas are more likely to be climbed by other lianas, because the existing lianas can provide structural support for access to the canopy (Putz 1984b; Perez-Salicrup et al. 2001).

. Liana abundance was significantly positively correlated with the tree diversity at both the reference rainforest and the edges in present study. This liana and tree association may be a result of response to rainforest disturbances. Both the diversity of trees and lianas can be promoted by disturbance (Connell 1978; Schnitzer et al. 2000; Laurance et al. 2001b; Molino & Sabatier 2001). For example, tropical cyclones may cause considerable destruction to rainforest which may lead to dramatic increases in abundance of lianas (Webb 1958; Laurance 1997). At the same time, disturbance may also increase the tree diversity following the establishment of shade-intolerant tree species (Vandermeer et al. 2000). Moreover, rainforest gaps have been found to be important in maintaining tree species diversity (Denslow 1987; Brokaw & Scheiner 1989; Brokaw & Busing 2000) as well as liana diversity (Babweteera et al. 2000; Schnitzer et al. 2000; Schnitzer & Carson 2001), thus the correlation of lianas with the diversity of trees may be a result of disturbance and successional dynamics, especially the elevated gap events at rainforest edges (Laurance et al. 1998a).

Lianas have been described as structural parasites which compete for resources with their host trees, both above-(light) and below- ground (water and nutrients) (Stevens 1987; Putz & Mooney 1991; Laurance et al. 2001b; Schnitzer et al. 2005). A negative correlation between tree biomass and liana abundance has frequently been found as lianas tend to slow the growth of trees, cause direct physical damage to their hosts, and increase the damage associated with tree falls (Gerwing & Farias 2000; Laurance et al. 2001b). This study, however, failed to find such a negative association between lianas and trees either at edges or the rainforest reference plot (represented by total tree basal area). Instead, the total liana basal area was positively correlated with the total tree basal area in the reference rainforest. Lianas tend to infest large trees which can provide them with access to the canopy (Perez-Salicrup et al. 2001). The subtropical rainforest, where the rainforest reference plot is located, is in a stable old 2 growth stage with very high total basal area of 70 mP P per ha (tree >10 cm)(Gentry 1991; Laidlaw et al. 2000). The longer exposure of large trees to liana infestation may have caused this liana and large tree association (Putz 1984a; Perez-Salicrup et al. 2001).

Positive correlations were also found between the density of lianas and the density, diversity and total basal area of small tree assemblages (DBH<10 cm) at rainforest edges. The density and diversity of trees has been found to increase after edge creation due to the establishment shade-intolerant species, responding to the increase in lateral light penetration and soil perturbations (Laurance et al. 1998a; Laurance et al. 2001b; Laurance et al. 2006b). In this study, the same result was found with large numbers of shade-intolerant species occurring near edges, such as Acronychia spp., Guioa semiglauca, H. kendack, Orites excelsus, Polyscias elegans, and Rhodomyrtus psidioides (Chapter 3). The abundance of early successional species of lianas, such as Cissus antarctica, Cissus sterculifolia, Cayratia eurynema, and Tetrastigma nitens (Hegarty 1989) also increased dramatically at edges. This implies that these liana and tree assemblages may respond to edge effects, especially increased light levels, similarly after a disturbance. In addition, forest with high abundance and diversity of trees may also be structurally attractive for lianas as newly established lianas need a series of trees of different sizes as support to access the canopy (Putz 1984b).

Most of the species that responded to the edges were fleshy-fruited - both trees and lianas. While most rainforest plants bear fleshy-fruits (Willson & Crome 1989), fleshy-fruits are particularly a characteristic of early successional species, promoting their dispersal by frugivores (Hopkins et al. 1977). Subsequently, the increase in fleshy-fruited species at rainforest edges may further increase frugivore visitations. In turn, this reciprocal relationship of fruigivore and fleshy-fruited plants may have profound impacts on the dynamics of the edges as well as the rainforest interiors in general.

Liana and tree associations have become one of the main foci in liana studies (Schnitzer & Bongers 2002; Putz 2004). A study in Bolivia showed that plots with similar tree species compositions did not have similar liana composition although there was an obvious association of some palm species with plots having high liana density (Perez-Salicrup et al. 2001). Many random ecological processes, such as gap dynamics, may be more important in determining the distribution of lianas in a forest (Schnitzer et al. 2000). The rainforest reference plot in the present study also did not show a clear liana and tree association in terms of their species composition. At the edges studied, however, strong tree and liana associations were found. Plots with

similar tree compositions were found to have similar liana compositions. This implies that the liana and tree associations may have been changed at rainforest edges due to elevated levels of disturbance and subsequent changes in environmental factors which determine the composition of plant assemblages. The similar responses of early successional lianas and shade-intolerant trees to the edge effects may be responsible for this liana and tree association at rainforest edges.

4.4.4 Change in liana climbing guilds

The changes in liana climbing guilds followed a gradient from edge to rainforest interiors and also varied from edge to edge, reflecting mainly the canopy openness and associated disturbances. The most significant change was the increase in both the density and proportion of tendril climbers at edges. Tendril-climbers have been shown to be a dominant life form in young secondary forests, and decrease with the increase in twiners as the forest stand ages (Dewalt 2000). These changes in liana composition over time reflect the changes in light regimes and the availability of host trees following forest succession (Dewalt 2000; Laurance et al. 2001b). The increase in assemblages of small trees at rainforest edges may also lead to an increase in tendril climbers, which tend to climb smaller trees (Putz & Holbrook 1991; Laurance et al. 2001b). An increase in abundance of tendril climbers was also found in Amazon rainforest edges but the proportion of different climbing guilds remained unchanged compared with rainforest interiors (Laurance et al. 2001b). This might be due to the fact that the Amazonian edges studies were relatively young (<20 yr), thus the proportions of different climbing guilds might strongly reflect the composition of lianas prior to the edge creation (Laurance et al. 2001b). The edges surveyed in the present study were over 50 years old. Interestingly, the proportion of tendril climbers at the edges is comparable to that of 20 year old secondary forest (Dewalt 2000), indicating the unstable state of rainforest edges.

4.4.5 Change in liana dispersal guilds

Significant increases in the proportion of frugivore-dispersed lianas were found at rainforest edges, compared with the rainforest reference plot, in which lianas were mainly wind-dispersed. These changes showed a gradient increasing from the

rainforest reference plot to roadside edges, pasture/ rainforest edges and peaking at the eucalypt forest/ rainforest edges. These changes were mainly contributed to by an increase in the abundance of a few native grapes such as C. antarctica, C. sterculifolia, C. eurynema, and T. nitens, which are considered early successional species (Hegarty & Caballé 1991). Similar patterns were evident in the density of shade-intolerant trees (Chapter 3). The same type of changes have also been found in the Amazonian rainforest fragments, in which the abundance of fleshy-fruited early successional tree species increased as much as 10 times at edges (Laurance et al. 2001b; Laurance et al. 2006b). This increase in frugivore-dispersed species at rainforest edges may substantially alter the dynamic of the rainforest interior, given the potential increase in frugivore activity in response to the increase in fleshy fruited plants around edges.

5 THE SEEDLING BANK AND RAINFOREST FRAGMENTATION

5.1 Introduction

A common phenomenon within a rainforest is the accumulation of tree seedlings beneath the dense canopy. This mass of immature rainforest trees is generally called the seedling bank (Whitmore 1996, 1998a; Marod et al. 2002). Composed of mainly mature phase rainforest species, the seedling bank makes up a large proportion of rainforest diversity and plays important roles in rainforest maintenance and dynamics (Garwood 1996; Richards 1996; Whitmore 1998a).

The seed of most mature phase rainforest species, in general, cannot persist in the soil seed bank as do many pioneer species (Garwood 1989; Vazquez-Yanes & Orozco- Segovia 1993). Accordingly, maintaining a reliable seedling bank is crucial for their successful recruitment as their reproductive cycle is generally long, seed production is largely unpredictable and seed predation is extremely high (Schupp et al. 1989; Whitmore 1989; Connell & Green 2000).

The inherent characteristics of each species, such as seed traits, germination and growth requirements, may decide whether or not it persists in the seedling bank (Foster & Janson 1985; Popma & Bongers 1988; Vazquez-Yanes & Orozco-Segovia 1993; Green & Juniper 2004; Gilbert et al. 2006). The composition of the seedling bank, however, is determined largely by the local forest stand, which provides direct seed sources as well as the environmental requirements for seed germination and seedling establishment and growth (Richards 1996; Webb & Peart 2000). Forest light- gaps play central roles in seedling bank dynamics and, therefore, contribute significantly to the maintenance of high diversity within rainforests (Denslow 1987; Brokaw & Scheiner 1989; Whitmore 1989; Brokaw & Busing 2000; Schnitzer & Carson 2001). Other ecological processes such as seed dispersal and predation, herbivory and pathogen infection also have profound impacts on the seedling bank at different stages from seed germination to seedling establishment (Augspurger 1984b;

Wang & Smith 2002; Benítez-Malvido & Lemus-Albor 2006). The mechanisms driving seedling bank dynamics and, in turn, its role in maintaining rainforest diversity have attracted much attention from ecologists. This has led to the development of several hypotheses regarding the origin and maintenance of rainforest diversity (Wright 2002), such as ‘Gap dynamics’ (Brokaw & Scheiner 1989), ‘Niche differentiation’ (Grubb 1977; Grubb 1996), ‘Distance and/ or density dependence’ (Janzen 1970; Connell 1971), ‘Compensatory recruitment’ (Connell et al. 1984) and ‘Recruitment limitation’ (Eriksson & Ehrlén 1992; Hubbell et al. 1999). Some of these hypotheses have received support from recent large scale rainforest demographic and dynamic studies (Hubbell et al. 1999; Harms et al. 2000).

Most of the seedlings in the seedling bank grow near their compensation point due to the limited light (less than 2%) reaching the forest floor after penetrating the multilayered canopy (Augspurger 1984a; Whitmore 1998a). Under these conditions, sun flecks become extremely important for the survival of seedlings under the deep shade (Chazdon 1988; Chazdon & Pearcy 1991; Whitmore 1996; Leakey et al. 2005). As a result, the current seedling distribution pattern may reflect largely the past changes in light regimes (Nicotra et al. 1999). For seedlings growing in the shade, survival is far more important than limited growth (Kitajima 1994). Some seedlings may persist in the seedling bank for a long time before a suitable light gap appears. At this time, the seedlings may then be ‘released’ and experience a period of rapid growth to approach the canopy (Brokaw 1987; Denslow 1987; Whitmore 1996; Dalling & Hubbell 2002).

Rainforest destruction and fragmentation have become one of the major threats to biodiversity globally (Whitmore 1997; Pimm & Brooks 1999; Sodhi et al. 2004; Brook et al. 2006; Wright & Muller-Landau 2006a). One of the direct consequences of rainforest fragmentation is the exposure of the seedling bank to edge effects following the changes in the forest canopy (Williams-Linera 1990a; Benitez-Malvido 1998; Laurance et al. 1998b). This may lead to the alteration of regeneration within rainforest remnants, affecting their long-term dynamics, and potentially causing direct extinction of many rare species (Hill & Curran 2001; Benítez-Malvido & Martínez- Ramos 2003a). Shortly after edge creation, the increased light regime can result in a short-term release of some seedlings in the seedling bank which then form a dense

layer of small trees along the edge (Williams-Linera 1990a; Sizer & Tanner 1999). Both the density and diversity of mature rainforest seedlings have been observed to decrease with the increase in secondary or weed species following rainforest fragmentation (Benítez-Malvido & Martínez-Ramos 2003a). These changes in seedling banks may be caused directly by the changes in physical environmental factors, such as the light regime, soil moisture and/ or temperature, or by modified ecological processes such as pollination, seed dispersal and herbivory (Lovejoy et al. 1984; Williams-Linera 1990a; Didham et al. 1996; Williams-Linera et al. 1998; Restrepo & Vargas 1999; Benítez-Malvido & Martínez-Ramos 2003b). Global climatic change, with predicted increase in cyclone events and rainfall seasonality, may also influence the seedling dynamics (Whitmore 1998b).

Most studies of rainforest fragmentation have focused on changes in the assemblages of large trees in remnants (Laurance et al. 1998a; Williams-Linera 2002). The effects of fragmentation on the dynamics of seedlings are poorly understood due largely to the difficulties in seedling identification (Benítez-Malvido & Martínez-Ramos 2003b). Moreover, liana seedlings are generally not included in the limited seedling studies which investigate rainforest fragmentation (Benitez-Malvido 1998; Sizer & Tanner 1999; Benitez-Malvido & Lemus-Albor 2005).

The subtropical rainforest in Lamington National Park provides a particularly attractive site for studies of the seedling bank as the species diversity is relatively low and seedlings are easier to identify compared with richer tropical rainforests. The presence of a mere 30 species of lianas makes it possible to include lianas in seedling studies. This allows the assessment of the potentially important relationship between lianas and trees at the seedling stage to be investigated.

In this study the following hypotheses have been tested: 1. the rainforest interior maintains a seedling bank with higher density and diversity of tree seedlings than the rainforest edges; 2. the density of shade-tolerant tree species in the seedling bank will decrease and that of shade-intolerant species will increase with proximity to rainforest edges; 3. there will be a higher density of liana seedlings near rainforest edges than within the rainforest interior; and

4. the composition of the seedling bank will be different at different type of edges.

5.1 Method

5.1.1 Seedling bank sampling at the one-hectare reference plot

Within the one-hectare plot, 50 of the 100 grid squares (odd numbered) were selected for the seedling study. In each grid square, five 1 × 1 m subplots were established for the seedling survey, with four at each of the corners and one at the centre of the grid square. All woody rainforest plants below 2.5 m were identified and the height measured within the subplots. Data from the five subplots were pooled to represent the seedling bank in the grid square. Meanwhile, trees ≥ 5 cm and lianas ≥ 2 cm (in DBH) were identified and measured at breast height (methods see chapter 2) for the analysis of correlations between the seedling bank and forest canopy,

5.1.1 Seedling bank sampling at rainforest edges

Nine 100 m transects were surveyed for the each of the three edge types (eucalypt forest/ rainforest, pasture/ rainforest and roadside/ rainforest). Along each 100 m transect, five 10 m ×10 m plots were established for the edge vegetation survey (methods see chapter 2). Within each of the 10 m × 10 m plot, five subplots were established following the same procedure as used in the one-hectare reference plot. Seedling data were collected in the subplots and then pooled to represent the seedling bank of the plot. Nine transects were surveyed for the each of the three edge types. In 2 total, 135 seedling samples (5×1 mP P each) were surveyed from the 27 transects across the three types of edges. In addition, the canopy trees (DBH≥ 5 cm) and canopy lianas (DBH≥ 2 cm) were identified and measured within each 10 m ×10 m plot for analysis of edge structures.

5.1.2 Data analysis

Seedlings were classified as either trees or lianas. Analyses were carried out for tree and liana seedlings separately.

The relative density of each tree species in the seedling bank and the associated

canopy trees was calculated using the number of seedlings or tree stems of a particular species divided by the total seedling or tree stems. The association of the seedling bank with the forest canopy was assessed by correlating the density and species richness of seedlings with the density, total basal area and species richness of associated canopy trees and lianas using Spearman Rank Correlation (Quinn & Keough 2002).

To assess the distribution of seedlings in different size class, a distribution curve was constructed using the number of seedlings in different size classes at 5 cm intervals in height. Different regression models were fitted using curve estimation in SPSS 13.0 for Windows (SPSS Inc., 2004).

The distributions of seedling densities at rainforest edges were compared using two- way ANOVA considering the distance from edge and edge types as fixed factors. The comparison between the rainforest reference plot and edges used one-way ANOVA with Tukey’s HSD to test the differences between the rainforest reference plot and each of the edge types. The distribution of dominant species in relation to their proximity to edges was assessed by correlating the seedling density with the distances of plots from the forest edge using Spearman Rank Correlation (Quinn & Keough 2002).

To compare the floristic composition in the seedling bank of trees among edges, a non-metric multi-scaling ordination (NMDS) was used based on Bray-Curtis similarity matrices calculated between transects, based on the abundance of species in the seedling banks. In addition, analysis of similarity (ANOSIM) was carried out to test the differences between edges. Both NMDS and ANOSIM were conducted using PRIMER 5 (Clarke & Warwick 2001). The associations of each tree species with the ordination coordinates were assessed using multiple regression using Pop Tools Excel add-in (Hood 2004), and their significance levels were further tested using a randomization test (Edgington 1980). For those species which showed a significant association with the ordination coordinates, their density at different positions from the edges was correlated with the distance from the edge to the interior using Pearson Correlation. Biplot vectors were generated using the correlation indices, showing the trends in increasing abundance within the ordination spaces.

5.2 Result

5.2.1 The seedling bank within the one-hectare reference plot

Seedling bank composition 2 In total 850 tree seedlings were identified within the 250 1 mP Pseedling plots. This gives an estimation of overall density of 17 ± 0.922 (mean ± SE) seedlings per square metre within the 1 ha plot. The seedlings represented 60 tree species (including three unidentified species). Thirty one species recorded in the tree survey had no seedlings present in the survey. These species, however, were also relatively rare in the forest, being mostly represented by only 1 or 2 stems within the entire hectare.

Table 5.1 Number of individuals and relative densities of the seedlings of the dominant species (with 20 or more seedling recorded) in the seedling bank and their relative densities in the tree canopy (DBH > 5cm) in the rainforest reference plot. Number of Relative density Relative density Species seedlings in seedlings in canopy trees Actephila lindleyi 228 0.268 0.124 Atractocarpus benthamianus 105 0.124 0.108 Stenocarpus salignus 72 0.085 0.005 Acmena smithii 40 0.047 0.003 Argyrodendron trifoliolatum 38 0.045 0.058 Arytera divaricata 29 0.034 0.002 Myrsine subsessilis 28 0.033 0.001 Argyrodendron actinophyllum 26 0.031 0.024 Dysoxylum fracerianum 24 0.028 0.006 Sarcopteryx stipata 23 0.027 0.013 Quintinia verdonii 21 0.024 0.009

Sub-canopy tree species, for example, Actephila lindleyi and Atractocarpus benthamianus, dominated the seedling bank (Table 5.1), composing 39 per cent of total seedlings , compared with 23 per cent of all tree stems (DBH >5cm). The relative densities of the dominant species in the seedling bank generally reflected their relative densities among the trees, with some exceptions (Table 5.1). In particular, Stenocarpus salignus, Acmena smithii, Arytera divaricata and Myrsine subsessilis were an order of magnitude more abundant (proportionally) in the seedling bank than as trees.

Nineteen species of liana were recorded in the survey, which included most of the liana species present in the forest canopy (DBH > 2cm). The density of liana seedling 2 was calculated as 11.32±1.5 (mean ± SE) per mP ,P which was significantly lower than the tree seedlings (Paired t-test, P < 0.001). Lianas from three species, Melodinus australis, Austrostenisia glabristyla and Parsonsia velutina made up 64 per cent of the total liana seedlings and 72 per cent of canopy lianas (DBH > 2 cm), respectively. Moreover, the dominant species in the seedling bank were also the most abundant species in the forest canopy. The relative density of each liana species in the seedling banks was comparable to its relative density among the canopy lianas (Table 5.2), with one exception: Morinda jasmenoides was an order of magnitude (proportionally) more abundant in the seedling bank than as canopy lianas.

Table 5.2 Number of individuals and relative densities of lianas in the seedling banks and their relative density in canopy lianas (DBH > 2cm) within the rainforest reference plot. Number of Relative density in Relative density seedling seedlings in canopy lianas Melodinus australis 86 0.357 0.389 Parsonsia velutina 50 0.207 0.148 Austrostenisia glabristyla 45 0.187 0.187 Morinda jasminoides 17 0.071 0.004 Palmera scandens 10 0.041 0.014 Parsonsia straminea 7 0.029 0.039 Hibbertia scandens 5 0.021 0.004 Cephalaralia cephalobotrys 5 0.021 0.007 Piper novae-hollandiae 4 0.017 0.007 Derris involuta 4 0.017 0.004 Caesalpinia subtropica 4 0.017 0.014 Celastrus celastroides 2 0.008 0.025 Embelia australis 2 0.008 0.035

Seedling and canopy associations The number and species of tree seedlings in the rainforest reference plot were significantly negatively correlated with the number of tree species within 10 m× 10 m subplots (Spearman Rank Correlation, P < 0.05, Table 5.3). The number of tree seedlings also showed a marginal negative correlation with the number of trees and the total basal area of lianas in the tree plots (P=0.082 and 0.080, respectively). The species diversity, total basal area of tree and the number and species of liana showed no correlation with tree seedlings.

The density of liana seedlings, however, showed a significant positive correlation with the diversity of trees (Table 5.3). No other variables from canopy trees and lianas showed a correlation with the liana seedlings.

Seedling size distribution The number of seedlings in different size classes was best fitted with a power 2 distribution (RP =0.786,P P<0.001), indicating that seedling survival was different among different size classes (Fig. 5.1). The densities of seedlings up to 25 cm in height exhibited a sharp decrease from small to higher size classes (Fig. 5.1). Above 25 cm class, the density of seedlings then followed a logarithmic distribution in 2 different size classes (RP =P 0.731, P<0.001).

Table 5.3 Correlations between tree and liana seedlings and canopy trees and lianas in the 1 ha rainforest reference plot (n = 50, significant results are highlighted, P<0.05). Tree seedling Liana seedling Density N. of species Density N. of species Number of trees -0.248 -0.191 -0.217 -0.194 Total basal area of trees -0.127 -0.150 0.120 -0.003 Tree species richness -0.329 -0.298 -0.239 -0.239 Tree diversity -0.182 -0.037 0.292 0.233 Number of lianas -0.174 -0.052 -0.100 -0.061 Total basal area of lianas -0.250 -0.115 -0.066 -0.059

Other life form A few of common shrub species were recorded frequently in the survey of the reference rainforest plot, including Triunia youngiana, Pittosporum multiflorum and the walking stick palm, Linospadix monostachya. Many of these shrubs were encountered as mature individuals. Occasionally, some seedlings from the genus Solanum were also recorded when the plots were located in or close to forest gaps.

5.1.1 The seedling bank at rainforest edges

Seedling bank composition One hundred and sixty eight species of tree seedlings were recorded in transects

located at the rainforest edge, all but four of which were identified to species (these four were identified to genus only). The remaining four species were identified to genus. The density of seedlings varied considerably among edge types (Table 5.4). Tree seedlings were mainly dominant canopy species, such as A. lindleyi, Pseudoweinmannia lachnocarpa, Argyrodendron actinophyllum, Elattostachys nervosa and Baloghia inophylla. Two species of small trees, Capparis arborea and Stenocarpus salignus, which were not common in the canopy, had many seedlings in the seedling bank. Around 80 species of tree seedlings were recorded across the three types of edges, but the eucalypt forest/ rainforest edges showed the highest species diversity in spite of the seedling density being the lowest.

15000

Eucalypt forest/rainforest edge 10000 Pasture/rainforest edge Roadside edge Reference rainforest

5000 Seedlings/ ha

0 0 20 40 60 80 100 120 140 160 180 200 220 240 260 Seedling size (cm)

Figure 5.1 Number of seedlings in different size classes at three rainforest edges and 2 within the rainforest reference plot. (45 pooled 5 mP seedlingP plots for each edge type 2 and 50 pooled 5 mP P seedling plots from the rainforest reference plot)

Nineteen species of liana were recorded, which were all found at roadside edges (Table 5.4). Liana seedlings were dominated by Derris involuta, which made up nearly 85% of all liana seedlings or 43% of all seedlings in the seedling bank. Because of that, the species diversity indices of liana seedling at edges were very low compared with that within the reference rainforest.

Distribution of seedling density The density of tree seedlings was significantly different across edges (Table 5.5). Post hoc tests showed that the densities of tree seedlings at roadside edges were

significantly higher than at eucalypt forest/ rainforest or pasture/ rainforest edges. There was no overall distance effect on the distribution of seedlings from the edge to forest interiors, but the significant interaction of edge type and distance effects indicated that the distribution patterns were different across different types of edges (Fig. 5.2).

Table 5.4 Densities (mean ± SE) and species richness of tree and liana seedlings at 2 rainforest edges. (45 pooled 5 mP P seedlingP plots for each edge type) Tree seedling Liana seedling

Edge type Density Species Diversity Density Species Diversity (per ha) richness (per ha) richness Eucalypt forest/ 22311 ± 1402 85 3.78 20489 ± 4728 11 0.86 rainforest Pasture/ 24088 ± 2657 74 3.57 22489 ± 5397 14 0.45 rainforest Roadside 31911 ± 2014 80 2.99 37022 ± 7395 19 0.87

At roadside edges, the density of tree seedlings were significantly negatively correlated with the distance from the edge to the rainforest interior (R=-0.348, P=0.019). In contrast, at the pasture/ rainforest edge, the densities of tree seedlings were positively correlated with distance from the edge, increasing towards the interior (R=0.366, P=0.013). There was no distance effect at eucalypt/ rainforest edges (R=0.011, P=0.945).

Table 5.5 Results of the two-way ANOVA for the seedling densities at different distances from edges to the rainforest interiors and among the rainforest edges All seedling Tree seedling Liana seedling Source Df F-ratio P-value F-ratio P-value F-ratio P-value Edge 2 6.708 0.002 6.439 0.002 2.241 0.111 Distance 4 0.059 0.994 0.427 0.789 0.343 0.848

Edge * Distance 8 0.614 0.765 2.553 0.013 0.730 0.664

The density of liana seedlings was similar across the three types of edges and showed no significant distance effects (Table 5.5, Fig. 5.2). Liana seedling density showed considerable variation among plots, indicating the patchy distribution of liana seedlings at edges.

60000 Eucalypt forest/rainforest edge Pasture/rainforest edge Roadside rainforest edge 40000

20000 Liana seedlings/ ha Tree seedlings/ Tree ha 0 0 20406080100Reference

100000

80000

60000

40000 Liana seedlings/ ha

20000

0 Reference 0 20406080100

Distances from edge to rainforest interior

Figure 5.2 Changes in the density of tree (above) and liana (below) seedlings from edges to rainforest interiors, with comparisons with rainforest reference plot. (9 2 2 pooled 5 mP seedlingP plots for each distance at edges and 50 pooled 5 mP P seedling plots from the rainforest reference plot; error bars represent the standard errors)

Comparison of seedling bank at edges with the rainforest reference plot The density of tree seedlings within the reference rainforest was similar to that at the roadside edge, but was significantly higher than that at eucalypt forest/ rainforest and pasture/ rainforest edges (One-way ANOVA, P < 0.001, Tukey’s HSD, Fig. 5.2). On the other hand, the density of liana seedlings within the reference rainforest was significantly lower than that at the roadside edges, but did not differ from either eucalypt forest/ rainforest or pasture/ rainforest edge (One-way ANOVA, p = 0.003, Tukey’s HSD, Fig. 5.2)

Comparisons of seedling bank composition among edges The MDS showed a good separation among edge types in terms of the tree seedling bank composition, with the roadside edges gathered in the centre of the ordination

space and eucalypt/ rainforest and pasture edges scattered in two different directions (Fig. 5.3). The ANOSIM result showed that the eucalypt/ rainforest and pasture/ rainforest edges were not significantly different from each other but both of them differed significantly from the roadside edge (Table 5.6).

1.5 Pasture/rainforest Roadside rainforest Eucalypt forest /rainforest

0.5

-0.5

-1 Stress = 0.17

-1.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 2

3 1 4 5 6 12 7 8 10 11 9 Figure 5.3 MDS results for the composition of tree seedlings among rainforest edges, with vectors showing species significantly associated with ordination axes. (Ellipse is drawn around all roadside edges). Species are: 1.Phaleria chermsideana 2.Pouteria australis 3.Capparis arborea 4.Daphnandra micrantha 5.Arytera distylis 6.Alectryon subcinerveus 7.Atractocarpus chartacea 8.Cleistanthus cunninghamii 9.Acmena smithii 10.Wikea macrophylla 11.Cinnamomum virens 12.Stenocarpus salignus

Table5.6 ANOSIM results comparing the composition of tree seedling banks among the rainforest edges (Global R =0.181, P=0.006) Source R Statistic P Pasture/ rainforest VS roadside edge 0.218 0.018 Pasture/ rainforest VS eucalypt forest/ rainforest edge 0.093 0.135 Roadside edge VS eucalypt forest/ rainforest edge 0.247 0.024

Most of the dominant species, such as A. lindleyi, P. lachnocarpa, A. actinophyllum, E. nervosa and B. inophylla, showed no association with any edge type, indicating a similarity in the seedling banks across edges. There were, however, 20 species which

showed significant associations with the MDS coordinates, of which 12 species had 10 or more seedlings recorded (Fig. 5. 3). Among these, Acmena smithii, Wilkea macrophylla, Cinnamomum virens and S. salignus were associated with roadside edges; Cleistanthus cunninghamii, Atractocarpus chartacea, Daphnandra micrantha, Arytera distylis and Alectryon subcinerveus were mainly associated with eucalypt forest/ rainforest edges; and Phaleria chermsideana, Pouteria australis and C. arborea showed associations with both eucalypt forest/ rainforest and pasture rainforest edges.

Table 5.7 Correlations between the densities of tree seedlings and the distances from edge to the rainforest interiors (only species with 20 or more seedlings are analysed). Number of Correlation P Seedling recorded coefficient Actephila lindleyi 401 -0.037 0.674 Pseudoweinmannia lachnocarpa 125 0.236 0.006 Elattostachys nervosa 82 -0.180 0.034 Argyrodendron actinophyllum 69 -0.211 0.014 Capparis arborea 62 0.133 0.125 Stenocarpus salignus 55 0.195 0.024 Baloghia inophylla 53 -0.102 0.240 Diospyros pentamera 42 -0.102 0.240 Dysoxylum fracerianum 41 -0.154 0.075 Argyrodendron trifoliolatum 38 0.046 0.599 Anthocarpus nitidula 38 0.234 0.006 Guioa semiglauca 35 -0.182 0.034 Daphnandra micrantha 33 -0.175 0.042 Atractocarpus benthamianus 33 0.096 0.267 Arytera divaricata 31 0.127 0.143 Sarcopteryx stipata 31 0.194 0.024 Arytera distylis 26 -0.102 0.240 Cleistanthus cunninghamii 22 0.158 0.068 Neolitsea australiensis 20 0.025 0.776 Myrsine subsessilis 20 0.095 0.275 Diploglottis australis 20 0.094 0.280

Different species exhibited different distribution patterns in response to distance from edge to rainforest interior (Table 5.7). The majority of species, such as the most abundant species, A. lindleyi, showed no response to distances. The seedling density of P. lachnocarpa, S. salignus, A. nitidula and S. stipata increased with the distance from the rainforest edge. In contrast, A. actinophyllum, E. nervosa, Guioa semiglauca and D. micrantha increased in densities towards the edges.

Seedling size class distribution The size class distribution of tree seedlings varied between edge types (Fig. 5.1). The distribution at roadside edges was best fitted with a power function, whereas those of eucalypt forest/ rainforest and pasture/ rainforest edges were best fitted with a logarithmic distribution (Table 5.8). Both eucalypt forest/ rainforest and pasture/ rainforest edges had a density of lower seedling densities in seedlings up to 20 cm than the roadside rainforest edge and the reference rainforest (Fig. 5.1)

Table 5.8 Comparisons of regression models of the distribution of seedlings in different size classes at rainforest edges and the rainforest reference plot.. Site type Model Model Summary Parameter Estimates R Square F Sig. Constant b1 Eucalypt forest/ rainforest edges Linear 0.472 34.03 <0.001 29.88 -0.17 Logarithmic 0.746 111.45 <0.001 75.90 -14.62 Power 0.399 25.18 <0.001 1441 -1.27 Logistic 0.491 36.70 <0.001 0.020 1.021 Pasture/ rainforest edges Linear 0.360 21.34 <0.001 32.98 -0.19 Logarithmic 0.631 64.95 <0.001 88.24 -17.21 Power 0.359 21.27 <0.001 2365.83 -1.41 Logistic 0.450 31.14 <0.001 0.018 1.02 Roadside/ rainforest edges Linear 0.414 26.89 <0.001 43.80 -0.26 Logarithmic 0.618 61.49 <0.001 109.31 -21.04 Power 0.663 74.74 <0.001 720.00 -0.98 Logistic 0.647 69.61 <0.001 0.02 1.02 Reference rainforest Linear 0.363 21.62 <0.001 60.93 -0.39 Logarithmic 0.730 102.70 <0.001 182.45 -36.96 Power 0.786 139.81 <0.001 1860.15 -1.21 Logistic 0.701 89.05 <0.001 0.018 1.02

Correlation between the seedling bank and the forest canopy The density of tree and liana seedlings showed different correlations with components of forest canopy at different edge types (Table 5.9). Canopy trees showed no relationships with seedling occurrence except that the total tree basal area was significantly positively correlated with liana species richness at pasture/ rainforest edges.

The number of stems, basal area and species richness of small trees were significantly negatively correlated with the density and species richness of tree seedlings at pasture/ rainforest edges (Table 5.9). There was no such correlation at the other two edge types.

The density of canopy liana was negatively correlated with tree seedlings at pasture/ rainforest edges but significantly positively correlated with tree seedlings and liana

species at roadside edges (Table 5.9).

There was no significant correlation between forest canopy and seedling banks at eucalypt forest/ rainforest edges.

Table 5.9 Correlations of the density and the number of species of seedlings with the stem density, basal area and species richness of canopy trees, small trees and lianas > 2cm at the rainforest edges (* P < 0.5, ** P < 0.01) Canopy tree (DBH>5cm) Small Tree ( DBH≤5cm) Liana (DBH≥2cm) Edge types and N. of Basal Species N. of Basal Species N. of seedling category stem area richness stem area richness Basal area stem Pasture/ rainforest edge Tree seedling Density -0.246 0.203 -0.098 -0.464** -0.421** -0.405** -0.285 -0.358* Species -0.188 0.237 -0.108 -0.429** -0.428** -0.349* -0.353* -0.389** Liana seedling Density 0.049 0.252 -0.013 -0.006 -0.119 -0.024 -0.048 0.009 Species 0.194 0.396** 0.112 -0.046 -0.136 -0.013 -0.021 0.028 Roadside/ rainforest edge Tree seedling Density 0.056 -0.060 -0.005 0.191 0.225 0.141 0.318* 0.393** Species 0.030 0.031 -0.031 0.200 0.229 0.230 0.341* 0.412** Liana seedling Density -0.102 0.160 -0.046 -0.154 -0.320 -0.144 0.083 -0.002 Species -0.014 0.191 0.070 0.088 0.007 0.112 0.389** 0.351* Eucalypt forest/ rainforest edge Tree seedling Density -0.008 -0.075 -0.023 -0.083 -0.046 -0.250 -0.144 -0.233 Species -0.125 -0.149 -0.072 -0.100 -0.175 -0.167 -0.084 -0.129 Liana seedling Density -0.174 0.004 0.010 0.103 -0.003 0.133 -0.042 0.150 Species -0.290 0.020 -0.078 0.009 -0.155 0.054 -0.006 0.073

Correlation between tree and liana seedlings In the one-hectare plot, the density of tree seedlings was much higher than the density of liana seedlings (Paired t-test, P<0.001). The density of tree and liana seedlings were significantly correlated with each other on 10 m × 10 m subplots (Pearson’s R= 0.368, P=0.009). At the edges, however, the density of tree seedlings did not differ from that of liana seedlings (Paired t-test, P=0.93) and associations between tree and liana seedlings varied between different edge types (Table 5.10). At both eucalypt forest/ rainforest and roadside/ rainforest edges, the density of tree seedlings was negatively correlated with the density of liana seedlings. There was no correlation between tree and liana seedlings at pasture/ rainforest edge.

Table 5.10 Correlations between the density of tree seedlings and liana seedlings in the rainforest reference plot and the rainforest edges. N of Correlation Site type plots coefficient Significance Reference rainforest 50 0.368 0.009 Eucalypt forest/ rainforest 45 -0.382 0.010 Roadside 45 -0.409 0.005 Pasture/ rainforest 45 0.049 0.747

5.2 Discussion

The results of this study suggest strongly that edge exposure has changed the composition of seedling banks along a gradient from edges to rainforest interiors and that these changes are affected substantially by edge types. According to the proximity to edges, some shade-tolerant species showed a decrease whereas some shade- intolerant species showed an increase in density of seedlings. The density of liana seedlings increased significantly at edges compared with the rainforest interior and this increase in liana seedlings may contribute to the decline of tree seedlings in two of the three edge types studied.

5.2.1 Matrix-regulated seedling banks

The results of this study are consistent with the literature that shows the impact of edge effects on seedling bank can vary dramatically depending on the type of edge and time since the edge was created (Kapos et al. 1997; Benitez-Malvido 1998; Sizer & Tanner 1999; Benítez-Malvido & Martínez-Ramos 2003b). The density of tree seedlings declined towards the pasture/ rainforest edges where they were significantly lower than densities in the reference rainforest. Similar patterns observed in Amazonian rainforest fragments surrounded by pastures after about 10 years were ascribed to a decrease in seed rain, and seed dispersal, and an increase in seed predation, resulting in lower seedling establishment (Benitez-Malvido 1998). The distribution of seedlings within different size classes in the present study suggests that decreased germination and/ or early seedling establishment may have contributed substantially to the lower density of tree seedlings near the edges. It is likely that herbivory may have played an important role in reducing seedling density near

pasture/ rainforest edges in this study. The herbivorous red-necked pademelons Thylogale thetis, use rainforest edges as their main feeding ground and their grazing may be responsible for the decrease in rainforest seedlings near edges (Wahungu et al. 2002).

At roadside edges, the density of tree seedlings showed the opposite distribution from that at the pasture/ rainforest edges, with an increase towards the edge. Winding through the dense primary forest, the section of the Lamington National Park road which was studied was gradually built from a horse track (Jackson & Stephenson 1986) and forest opening has been minimized by avoiding removal of big trees. It is likely that the regeneration of the roadside edges is similar to that in forest light gaps with limited canopy opening and relatively unchanged forest structure. Increases in seedling density at the edges were most likely caused by elevated light levels promoting the germination and establishment of rainforest seedlings. While roads are well known as corridors through which exotic species invade newly exposed environments (Forman & Alexander 1998; Goosem & Turton 2006), surprisingly, there was no obvious sign of weed infestation along the road studied although much more open areas around O’Reillys guest house were invaded by various weed species, such as wild tobacco Solanum mauritianum and Lantana camara (Hopkins 1975). This result supports the practice of minimizing the canopy opening during road construction through conserved forest.

The density of tree seedlings did not vary across the eucalypt forest/ rainforest edges however tree seedling density was significantly lower than in the reference rainforest. The dense shrub and small tree layer at this kind of open forest / rainforest edge may prevent lateral and vertical light penetration, thus maintaining a very low light level in the understorey (Unwin 1989; Turton & Freiburger 1997; Nicotra et al. 1999). This limited understorey light may be low enough to prevent the germination of some pioneer species (Turton & Duff 1992). Moreover, frequent fire events along the edge may also promote the dense shrub layer and cause seedling death directly (Smith & Guyer 1983; Unwin 1989). Unlike those forest edges in the Amazon which are surrounded by secondary rainforest that can act as a seed resource for edge regeneration (Sizer & Tanner 1999; Benítez-Malvido & Martínez-Ramos 2003b; Laurance et al. 2006b), the eucalypt forest/ rainforest edge is surrounded by open

forest composed of species that are almost exclusively confined to eucalypt forest (Unwin 1989; Bowman 2000). Frequent disturbance, such as fire and cattle grazing, plays an important role in maintaining this abrupt rainforest edge, and may also have a profound impact on the dynamics of tree seedling bank (Unwin 1989; Bowman & Panton 1993; Bowman 2000; Russell-Smith et al. 2004b)

Because so many factors can interfere with the magnitude and distance of the influence of the edges, it has been suggested that the vegetation responses to edges may be site specific (Harper et al. 2005). Many edge effects tend to become less obvious as the edges age and become more stable (Kapos et al. 1997). The composition of the seedling bank at older edges may be controlled more by the local disturbance regime than by the proximity to the edges (Lawes et al. 2005). This study has shown that the surrounding matrix and associated disturbances may well play a central role in regulating the dynamics of rainforest edges.

5.2.2 Seedling banks and the forest canopy

Assuming a similar species pool, the most important factor regulating the local seedling bank is the light environment imposed by the understorey (Augspurger 1984a; Macdougall & Kellman 1992; Whitmore 1996; Montgomery 2004). Montgomery and Chazdon (2001) found that the light regime in the forest understorey was largely determined by the sapling and shrub density, rather than the local canopy tree density or basal area. The present study also showed that the composition and structure of local canopy trees was a poor indicator of local seedling density.

In the one-hectare rainforest reference plot, only the number of tree species showed a significant but negative correlation with both the density and species richness of tree seedlings. This relationship is likely to be influenced by the disturbance regimes within the rainforest. Many studies have shown that disturbances may promote local tree species diversity (Connell 1978; Brokaw & Scheiner 1989; Molino & Sabatier 2001; Schnitzer & Carson 2001). At the same time, subsequent environment changes following disturbance may lead to the release of suppressed seedling banks (Denslow 1987; Popma & Bongers 1988; Whitmore 1997).The increase in physical damages may also directly cause seedling death, thus reducing the local seedling density (Clark

& Clark 1989; Nadolny 1999; Scariot 2000). On the other hand, disturbance may also promote the germination and establishment of seedlings which could lead to an increase in seedling density (Augspurger 1984b; Brokaw 1987; Vazquez-Yanes & Orozco-Segovia 1993; Dalling et al. 1998a). The negative relationship between canopy tree and seedling may be related to the magnitude and time of disturbance with, possibly, other ecological processes involved.

The negative correlation of the small tree assemblage and the density of tree seedlings at pasture/ rainforest edges may reflect the light regime beneath the forest understorey which is regulated by the small tree assemblages, as suggested by Montgomery and Chazdon (2001). Canopy lianas were also negatively correlated with tree seedling abundance at this edge type. Lianas are known to cause negative impacts on the growth and survival of trees by competing from both above and below ground, even from the seedling stage (Stevens 1987; Laurance et al. 2001b; Schnitzer & Bongers 2002; Schnitzer et al. 2005). The high rate of infestation of lianas near edges may reduce the understorey light regime thus inhibiting seed germination and seedling establishment. The negative correlation of liana and seedling density at pasture/ rainforest edges, however, may be related also to other factors, such as increased light promoting liana infestation (Laurance et al. 2001b) and increased herbivory reducing seeding density (Wahungu et al. 2002), because these associations were not consistent across edge types.

The species composition of the seedling bank, however, showed a close relationship with the composition of the canopies, which act as a direct seed source and maintain many local ecological processes regulating the dynamics of seedling bank.

5.2.3 Response of tree seedling to edge effects

Based principally on the requirements of seed germination and seedling establishment, Swaine and Whitmore (1988) classified rainforest trees into two major groups: pioneer species and climax species. The recognition of these two ecological groups has been widely used in studies of rainforest fragmentation, especially those that focus on vegetation responses to edge effects (Benitez-Malvido 1998; Laurance et al. 1998b; Mesquita et al. 1999; Sizer & Tanner 1999; Laurance et al. 2006b). Most

edges in the present study are more than 50 years old (Hopkins 1975) and have already been well ‘sealed’. Seedlings of pioneer species are not common in the seedling bank, in contrast with studies of newly created edges, for example those in Amazonian rainforest fragments (Kapos et al. 1997; Sizer & Tanner 1999). Instead, the seedling bank is composed mainly of mature-phase rainforest seedlings which exhibited different distribution patterns from edge to rainforest interiors.

On the basis of the seedling distributions, three types of responses to edge proximity can be identified. The majority of species showed no significant response to the distance effect, some species showed positive responses, whereas yet others showed negative responses. Comparable patterns were found in patches of tropical riparian forest in which the understorey light regime was suggested to be the driving force determining seedling distributions (Macdougall & Kellman 1992). In the present study, species with higher seedling densities near edges have all been reported to be ‘shade-intolerant’, such as G. semiglauca, or ‘intermediate’, such as A. actinophyllum and E. nervosa (Shugart et al. 1980). This demonstrates the importance of light in regulating edge dynamics. On the other hand, those species with lower seedling densities, such as P. lachnocarpa, S. salignus, A. nitidula and S. stipata, are all shade- tolerant species. The low density of those shade-tolerant species near edges may be caused by increased pathogen damage (Benitez-Malvido & Lemus-Albor 2005), lower seed germination caused by unfavourable environmental factors (Bruna 2002) and increased herbivory (Wahungu et al. 2002). For P. lanchnoarpa, low germination rate, possibly caused by low humidity, is likely to be the cause of lower seedling density near edges, as many un-germinated seeds were frequently found. In addition, seed dispersal limitation caused by a decrease in the number of dispersal agents, especially those that disperse large rainforest seeds, may significantly lower the density of rainforest seedlings near edges (Wright et al. 2000; Terborgh & Nunez- Iturri 2006).

The current seedling distributions are likely to be the result of interactions of many environmental factors and ecological processes. The increase in herbivory at rainforest edges may contribute substantially to the decline in overall tree seedlings at edges. For example, red-necked pademelons Thylogale thetis have been found regularly using the pasture/ rainforest edge as a feeding zone in the study area

(Wahungu et al. 2002). In that study, the damage to seedlings by this macropodid marsupial was significantly correlated with the distance from edges including for the tree species, A. actinophyllum. In the present study, however, the seedling density of A. actinophyllum increased near edges. It seems that the advantage from the increased light environment might have compensated for the negative effect from increased herbivory near the edge.

Edge effects have been widely accepted to be deleterious for rainforest fragments, causing continuing degradation with a consequent compromise of the conservation value of remaining remnants (Janzen 1970; Murcia 1995; Laurance 1997; Laurance et al. 2002), Nevertheless, this study shows that some rainforest tree species can benefit from edge effects with increasing seedling establishment near edges.

5.2.4 Liana and tree seedling associations

Lianas are well known to adapt to disturbance and generally increase in abundance and diversity following rainforest fragmentation (Laurance et al. 2001b; Schnitzer & Bongers 2002). Lianas and trees compete with each other not only for above ground light, but also for nutrients and water below ground (Schnitzer et al. 2005). The results from the present study show that competition between trees and lianas may start early in seedling stages at rainforest edges. The changed forest structure may favour the germination and establishment of some liana species, thus increasing the dominance of lianas at rainforest edges.

Liana seedlings make up about one fifth of all woody seedlings in a tropical forest understorey (Putz 1984b; Putz & Chai 1987; Wright et al. 2004). Within the reference rainforest in this study, the density of liana seedlings was much lower than that of the tree seedlings, making up 23% of the total woody seedlings. Further, the abundances of liana and tree seedlings were positively correlated with each other. This generally reflected the heterogeneity and patchy distribution of conditions suitable for seed germination and establishment of any kind. Current seedling distributions may reflect past patterns of light distribution within the forest (Denslow 1987; Nicotra et al. 1999). Natural disturbance, especially the occurrence of light gaps, may contribute to this distribution by promoting seed germination and increasing the seedling survivorship

of trees as well as lianas (Augspurger 1984b; Denslow 1987; Schnitzer & Bongers 2002; Gehring et al. 2004).

At rainforest edges, the density of liana seedlings increased dramatically (to around 50% of woody seedlings) to a degree that may potentially reduce the germination and the likelihood of survival of rainforest trees. This increase in liana seedlings may reduce the light reaching the forest floor, thus reducing the number of suitable sites for subsequent germination of other species. At the same time, below ground competition between liana and tree seedlings may affect the establishment of tree seedlings (Lewis 2000; Schnitzer et al. 2005). This competition may contribute substantially to the decline of tree seedlings near edges. Nicotra et al. (1999) found that second-growth forests tended to have a higher frequency of micro-sites with intermediate light level leading to a reduction in the spatial heterogeneity of light environments presented by old-growth forests. It is likely that lianas have benefited from, to some extent, the homogenized microenvironment at rainforest edges which may promote their dominance in the seedling bank. The dominant liana seedling, D. involuta, retains a pair of large first true leaves (aoround 40 × 30 mm) which can maintain photosynthesis under edge conditions leading, presumably, to enhanced survival. Surprisingly, D. involuta was not common in the canopy with only one stem in the whole one-hectare reference plot and few stems at the edges recorded among the canopy lianas (DBH>2 cm). Possibly a combination of edge effects and mast seeding events may have produced the even-aged seedling cohort.

The study area has been subjected to a prolonged drought with annual rainfall below average since 2000 (Chapter 2, Fig. 2.2). Moreover, the annual rainfall was only 707 mm in 2002 which was the driest year ever recorded (less than half of the average of 1601mm). Prolonged drought can change the phenological behavior of many rainforest trees causing synchronized mast fruiting (Ashton et al. 1988; Keeley & Bond 1999; Williamson & Ickes 2002). This mast fruiting can generate large cohorts of rainforest seedlings, which has been proposed as an evolutionary strategy to satiate natural enemies, both seed predator and seedling herbivores (Janzen 1974; Whitmore 1998a; Curran & Webb 2000; Williamson & Ickes 2002). The high density of D. involuta seedlings in the present study may be produced by a mast fruiting induced by the prolonged drought.

100

Moreover, the drought may increase the level of light reaching the forest understorey due to elevated death of trees (Nakagawa et al. 2000; Laurance et al. 2001a) and defoliation in the forest (Becker & Smith 1990), thereby increasing the survival of seedlings that can cope with the water stress, e.g. tap-rooted lianas (Aide & Zimmerman 1990; Schnitzer 2005). Seedlings of D. involuta form a typical tap root which may allow them to cope with the water stress and, at the same time, take advantage of increased understorey light through the pair of large first leaves.

Prolonged drought may also cause an extremely high death rate of newly established seedlings due to water stress (Gilbert et al. 2001; Delissio & Primack 2003; Bebber et al. 2004). It may also lead to changes in many ecological processes such as herbivory and seed predation, which, in turn, affect the composition of seedling banks. The prolonged drought in the study area may have had a strong impact on the seedling bank and this may be amplified by an interaction of the drought and edge effects.

The negative correlation of liana and tree seedlings seen at eucalypt forest/ rainforest and roadside edges indicates that the increase in liana seedlings may have reduced the survival of tree seedlings at edges. This increase in liana seedlings may lead to prolonged impacts on the dynamics of rainforest edges. Long-term monitoring is needed to assess the potential effects on tree regeneration.

101

102

6 THE SPATIAL DISTRIBUTION OF THE SOIL SEED BANK AND CANOPY TREES IN A SUBTROPICAL RAINFOREST

6.1 Introduction

Soil seed banks are composed of viable seeds in soil and associated leaf litters and play important roles in rainforest dynamics and regeneration (Whitmore 1983; Garwood 1989). The majority of seeds in primary rainforest soil seed banks are of pioneer species which need forest gaps to initiate germination and provide resources for seedling establishment (Symington 1933; Cheke et al. 1979; Hopkins & Graham 1983; Whitmore 1983; Garwood 1989; Dalling et al. 1997). Seedlings of these pioneer species are generally absent under the intact rainforest canopy due to insufficient light for seed germination or seedling establishment (Cheke et al. 1979; Putz 1983; Whitmore 1983; Garwood 1989). As a result, the species composition of soil seed banks differs substantially from that of the mature rainforest canopy, which is composed mostly of species that have developed a fast germination strategy and are capable of regenerating beneath a dense canopy (Hopkins & Graham 1987; Garwood 1989; Vazquez-Yanes & Orozco-Segovia 1993; Whitmore 1996).

The seed density and species composition of the soil seed bank varies dramatically among different types of forests and also within the same forest , both spatially and temporally (Garwood 1989; Dalling et al. 1997; Butler 1998; Dalling et al. 1998b; Singhakumara et al. 2000). Local disturbances, such as forest gap formation and anthropogenic (logging, slashing, burning), can have profound impacts in determining this variation (Putz 1983; Saulei & Swaine 1988; Hopkins et al. 1990a; Cao et al. 2000; Miller 2000). The abundance of seeds in soil seed banks has been found to be randomly distributed despite some species having clumped seed accumulations locally (Butler 1998).

At a local scale, the seed density of tropical pioneer tree species in soil seed banks decreases linearly away from their parent trees. At a broader sale, however, the abundance of seeds does not correlate with the density of reproductive-sized adult

103

trees in a 50 hectare plot on Barro Colorado Island (Dalling et al. 1998a; Dalling et al. 2002). The spatial distributions of seeds in soil seed banks, especially in relation to the distribution of reproductive trees, are still largely unknown (Garwood 1989; Dalling et al. 1998a), mainly due to spatial and temporal separation of seeds and trees of species represented in the soil seed bank (Saulei & Swaine 1988; Garwood 1989; Dalling et al. 1998b).

Compared with many Neotropical rainforest, one of the significant characteristics of Australian subtropical rainforest is the large accumulation of seeds from dominant canopy tree species in the soil seed bank, such as Caldcluvia paniculosa, Ceratopetalum apetalum, Geissois benthamii and Pseduoweimania lanchnocapa (Abdulhadi 1989; Floyd 1990; Olsen 1990). This phenomenon provides an opportunity to investigate the reciprocal relationship of the spatial distribution of the soil seed bank and seed trees, which is needed to understand the underlying determinants of recruitment patterns of rainforest trees.

In this chapter, the composition and spatial distribution of the soil seed bank and its relationship with above ground vegetations has been analyzed in a 1 ha plot within a subtropical rainforest. The specific questions addressed were:

1) How are seeds distributed spatially within the rainforest? 2) How does the structure of above ground vegetation affect the distribution of the soil seed bank; and 3) How does the distribution of conspecific parent trees affect the distribution of seeds in the soil seed bank?

6.2 Methods

This study was carried out within the 1 ha rainforest reference plot (for site location and description see Chapter 2), where 10 cm × 10 cm square samples of the top 5 cm of soil were collected from the center of each of the 100 grid-squares and germinated in a shade house (for sampling and germination methods see Chapter 2). The distribution of these germinated seedlings was then compared to the distribution of above ground vegetation including trees, lianas and seedlings in the 10 × 10 m plots.

104

Detailed vegetation surveys in each of the one hundred grid-squares provided the data on trees and lianas (Trees ≥ 5 cm, lianas ≥ 2 cm, Chapter 2), while seedling data were obtained from the seedling survey conducted in half (50) of the grid-squares in the 1 ha plot (for methods, see Chapter 2).

The species richness in the soil seed bank was estimated by comparing the species accumulation curve derived from 100 randomized species recordings with a first- order jackknife species richness estimation generated with EstamateS 7.5 (Colwell 2005). The species diversity in the soil seed bank was calculated using the Shannon Index (Magurran 2004).

To test whether or not the species composition and structure of above-ground vegetation could affect the seed distribution in the soil seed bank, the seed abundance and number of species germinated from each soil sample was correlated with trees, lianas and seedling in the associated tree plot using Spearman Rank Correlations (Quinn & Keough 2002).

The spatial autocorrelation of seed density was assessed using Moran’s I statistic at a distance lag of 10 metres (The approximate minimum distance between sampling spots). The significance of Moran’s I statistic was then tested using 5000 random perturbations in the program AutocorQ v.2.00 (Hardy & Vekemans).

The dominant species in the soil seed bank, C. paniculosa, also had many canopy trees in this 1 ha plot. This provides an opportunity to investigate the distribution of seeds in the soil seed bank in relation to the distribution of mature sized seed trees (DBH>30 cm). The distance from sampling spots to the nearest mature sized tree of C. paniculosa were calculated and then correlated with the number of seed germinated from each soil sample using Spearman Rank Correlation (Quinn & Keough 2002).

6.3 Results

6.3.1 Summary of the germination experiment

In total, 918 seeds were germinated from the 100 soil samples (Table 6.1). The

105

number of seeds germinated from each of the soil samples varied from 0 to 69, indicating the high spatial variation in seed distribution within the 1 ha plot.

2 Table 6.1 Seed densities (seed/ mP )P and number of species in the soil seed bank of the 1 ha subtropical rainforest plot, based on the germination of 100 samples each 10 cm × 10 cm and 5 cm deep. Total Tree Shrub Vine Herb Grass Total Seeds 918 644 139 21 109 5 Native 820 603 135 21 57 3 Exotic 98 41 4 0 52 2 Total Species 59 19 8 11 18 3 Native 46 18 7 11 8 2 Exotic 13 1 1 0 10 1

Fifty nine species were recorded in the germinated seed bank; 13 of which were exotic (Table 6.1). The average number of species germinated from each soil sample was 3.6 (range from 0 to 8 species). The species accumulation curve did not reach an asymptote which implies that more species may be expected if more samples were collected (Fig. 6.1). The first-order jackknife estimation suggests that 83 species may be expected in the soil seed bank. The value of the Shannon index was 2.06, which was rather low owing to the dominance of C. paniculosa.

Tree seeds were most numerous in the soil seed bank, comprising 70% of total germinated seedlings. Nearly 80% of tree seeds, however, were contributed by C. paniculosa (Cunoniaceae) alone. Herbs were also a common life-form in this soil seed bank, contributing nearly as many species as trees. Herbaceous seeds, however, only contributed 12% of the total seed germinated. Shrub seeds made up 15% of seeds in the soil seed bank and was dominated by a single species, Rubus rosifolius (Rosaceae), which made up 60% of the total shrub seeds germinated. Grasses were not common in the soil seed bank with only 3 species and 5 individuals.

Native seeds dominated the overall soil seed bank, accounting for 90 percent of the germinated seedlings (Table 6.1). Exotic seed was composed of a small tree species, Solanum mauritianum (Solanaceae), and many annuals especially Oxalis corniculata (Oxalidaceae) and those of Asteraceae such as Conyza canadensis (Table 6.2). The 11 vine species recorded were all natives.

106

90 80

70 ecies p 60

50 40 30 Observed species richness 20 10 First-order jackknife estimate

Cumulative number of s 0 0 20406080100 Cumulative number of samples

Figure 6.1 Species accumulation curve based on the observed species richness in the soil seed bank of the 1 ha subtropical rainforest plot, fitted with a first-order jackknife estimate. Based on the germination of 100 samples each 10 cm × 10 cm and 5 cm deep

Table 6.2 Number of seeds of dominant species germinated from 100 soil samples within the 1 ha subtropical rainforest plot (species with 10 or more seedling germinated are listed) Germinates Species Family Life form Native/ exotic (% of total) Caldcluvia paniculosa Cunoniaceae Tree Native 510 (0.56) Rubus rosifolius Rosaceae Shrub Native 85 (0.09) Hydrocotyle pedicellosa Apiaceae Forb Native 43 (0.05) Solanum mauritianum Solanaceae Tree Exotic 41 (0.04) Solanum aviculare Solanaceae Shrub Native 26 (0.03) Pseudoweinmannia lachnocarpa Cunoniaceae Tree Native 21 (0.02) Oxalis corniculata Oxalidaceae Forb Exotic 21 (0.02) Dendrocnide excelsa Urticaceae Tree Native 19 (0.02) Duboisia myoporoides Solanaceae Tree Native 17 (0.02) Solanum semiarmatum Solanaceae Shrub Native 15 (0.02) Conyza canadensis Asteraceae Forb Exotic 10 (0.01)

6.3.2 Soil seed bank and above-ground vegetation

Within the 1 ha plots, there were 10 species in the soil seed bank which also had at least one canopy tree in the forest (DBH>5cm, Table 6.3). Those species presented in both the soil seed bank and canopy, comprised 86% of total tree seeds in the soil seed bank and 22% of total individuals, or 32% of the total basal area of canopy trees. Seventy-six out of 86 canopy tree species were not recorded in the soil seed bank.

Nine species were present in both the soil seed bank and seedling bank (Table 6.3)

107

and these nine species contributed 95% of the total tree seeds in the soil seed bank and 16% percent of the seedlings. One species, Atractocarpus benthamianus had many newly germinated seedlings (0.4 seedlings per square metre) during the seedling survey. This species has been considered as a climax species (Shugart et al. 1980) and is generally absent from soil seed banks (Abdulhadi 1989; Stewart 1995). Only two seeds were germinated from soil samples. If this species was excluded from the soil seed bank, the proportion of seedlings of the shared species would be only 6.9% of total seedlings.

Table 6.3 The occurrence of species in the soil seed bank which were also present as trees or as seedlings within the 1 ha rainforest plot. Species Number of records Tree Seed Seedling Acacia melanoxylon 1 1 0 Acronychia laevis 4 4 6 Acronychia suberosa 31 7 3 Alangium villosum 1 2 0 Atractocarpus benthamianus 149 2 105 Caldcluvia paniculosa 69 510 3 Dendrocnide excelsa 0 19 5 Ficus watkinsiana 5 1 0 Polyscias elegans 5 4 8 Pseudoweinmannia lachnocarpa 35 21 1 Rhodomyrtus psidioides 2 2 4 Solanum mauritianum 0 41 6

The above-ground vegetation, including trees, lianas and seedlings, appeared to have only a small impact on the overall abundance of seed in the soil seed bank, at the scale of the 10m × 10 m plots (Table 6.4). The number of canopy lianas showed a significant but negative correlation with the density of non-woody seeds (Spearman Rank Correlation, R = -0.279, P = 0.004) and the density of exotic seed (Spearman Rank Correlation, R = -0.279 P = 0.012). The density of non-woody seeds was significantly positively correlated with the density of exotic seeds (Spearman Rank Correlation, R = 0.512, P < 0.001) as nearly half of the non-woody seeds were also exotic. The number of tree species also showed a negative correlation with density of non-woody seeds in the soil seed bank (Spearman Rank Correlation, R = -0.251, p = 0.021).

108

The total number of seedlings in the seedling bank, including tree and liana seedlings, was negatively correlated with the density of woody seeds in the soil seed bank (Spearman Rank Correlation, R = -0.285, P= 0.046).

Table 6.4 Spearman Rank Correlations between variables from the soil seed bank and those from the above-ground vegetation in 10 m × 10 m plots of the 1 ha rainforest reference plot (significant results are highlighted, P<0.05) Total Seed Non-woody woody Native Exotic Seed species seed seed seed seed Forest canopy (N=100) Tree stem 0.076 -0.025 -0.089 0.106 0.093 -0.011 Tree species 0.059 -0.130 -0.251 0.134 0.118 -0.196 Total tree basal area 0.190 0.067 -0.047 0.167 0.153 0.086 Tree diversity 0.059 -0.051 -0.073 0.119 0.126 -0.120 No. of canopy liana -0.034 -0.075 -0.279 0.048 0.037 -0.279 Seedling (N=50) Species of seedling -0.257 -0.023 0.054 -0.237 -0.237 -0.069 No. of all seedling -0.196 0.042 0.177 -0.285 -0.253 0.058 No. of liana seedling -0.188 -0.142 -0.047 -0.214 -0.162 -0.132 No. of tree seedling -0.142 0.172 0.264 -0.218 -0.222 0.198

6.3.3 Spatial distribution of seeds in the soil seed bank

The total number of seeds germinated from each of the 100 soil samples showed significant spatial autocorrelation at distances of 10 m, 20m, and 90m (Table 6.5). This pattern, however, was mainly caused by seeds of C. paniculosa which showed significant spatial pattern at almost all distances from 10 to 100 m except 50 and 60 m. If the seeds of C. paniculosa are excluded, the seed abundance showed almost no spatial autocorrelation except at the distance of 80 m. Among the five other species with ten or more seeds recorded, three of them, Hydrocotyle pedicellosa (Apiaceae), R. rosifolius and S. mauritianum showed no spatial autocorrelation, whereas two other species, S. aviculare and P. lachnocarpa, showed significant autocorrelation at various distance (Table 6.5). The distribution of seeds of H. pedicellosa, R. rosifolius, S. mauritianum and S. aviculare and P. lachnocarpa are shown in Figure 6.2.

109

110 Table 6.5 Moran’s I statistic on the spatial distribution of seeds in the soil seed bank of the 1 ha subtropical rainforest plot, based on 100 soil samples (10cm × 10 cm area, 5 cm deep) (significant autocorrelations are highlighted) 10 m 20m 30m 40m 50m 60m 70 80 90 100 Total seed I 0.207 0.133 0.051 0.085 0.029 -0.018 -0.065 -0.083 -0.148 -0.066 P 0.005 0.027 0.179 0.023 0.421 0.662 0.105 0.039 0.003 0.237 Seeds without Caldcluvia. I 0.111 0.089 -0.010 0.026 0.054 -0.008 -0.084 -0.110 -0.013 -0.035 paniculosa P 0.141 0.087 0.830 0.481 0.117 0.912 0.057 0.005 0.802 0.505 Caldcluvia paniculosa I 0.329 0.246 0.178 0.169 0.056 -0.061 -0.102 -0.154 -0.262 -0.296 P <0.001 <0.001 <0.001 <0.001 0.089 0.095 0.009 <0.001 <0.001 <0.001 Hydrocotyle pedicellosa I -0.096 0.046 0.031 0.046 -0.011 -0.006 0.011 -0.011 -0.016 -0.043 P 0.151 0.347 0.405 0.205 0.738 0.882 0.724 0.806 0.692 0.393 Rubus rosifolius I -0.099 0.051 -0.060 0.036 0.017 -0.030 0.001 0.018 0.024 0.024 P 0.181 0.349 0.103 0.331 0.612 0.435 0.978 0.596 0.592 0.658 Solanum mauritianum I 0.019 -0.030 0.035 -0.023 0.045 0.013 -0.043 -0.048 0.000 0.049 P 0.706 0.602 0.341 0.517 0.135 0.644 0.233 0.163 0.998 0.285 Solanum aviculare I -0.054 -0.009 0.108 -0.003 -0.024 -0.006 0.003 -0.004 -0.102 0.115 P 0.449 0.942 0.013 0.954 0.433 0.850 0.874 0.930 0.005 0.063 Pseudoweinmannia lachnocarpa I 0.168 0.253 0.053 0.052 -0.013 -0.005 -0.058 -0.085 -0.083 -0.130

P 0.065 <0.001 0.175 0.151 0.756 0.984 0.029 <0.001 0.017 0.029

Table 6.6 Moran’s I and P-value of significance tests from a spatial autocorrelation analysis of the canopy trees of C. paniculosa in the 1 ha subtropical rainforest pot, based on their x y coordinates within the 1 ha plot (Significant autocorrelation showed with bold numbers) Size classes (in DBH) 10m 20m 30m 40m 50m 60m 70m 80m 90m 100m 5 -15 cm I -0.003 0.021 -0.031 -0.043 0.027 0.007 -0.012 0.007 0.036 -0.017 P 0.986 0.690 0.449 0.253 0.415 0.874 0.814 0.822 0.421 0.712 15-30 cm I -0.007 0.025 0.007 0.009 0.001 -0.017 0.001 0.018 -0.048 0.007 P 0.973 0.608 0.822 0.776 0.994 0.622 0.974 0.632 0.247 0.904 30cm and above I 0.102 0.163 0.109 0.098 -0.068 0.018 -0.076 -0.113 -0.111 -0.045 P 0.182 0.007 0.015 0.009 0.033 0.656 0.043 0.007 0.011 0.427

100 100 2 1 1 5 1 90 90 2 1 2 4 1 2 2 80 80 1 2 2 1 2 1 70 70 1 4 2 1 1 1 1 60 60 1 1 1 1 2 3 50 50 1 5 1 2 2 1 1 2 2 40 40 1 2 1 2 4 2 1 30 30 2 1 1 3 3 20 20 1 1 22 2 1 11 1 1 10 10 1 331 2 1 1 1 1 4 1 2 1 0 0 0 20406080100 0 20406080100 Rubus rosifolius Hydrocotyle pedicellosa

100 100 1 1 1 1 6 44 90 90 1 1 1 1 80 80 1 1 1 1 1 1 70 70 2 1 3 1 60 60 2 2 1 50 50 1 1 121 1 40 40 2 1 30 30 3 2 1 20 20 1 1 10 10 3 1 1 1 1 2 0 0 0 20406080100 0 20406080100

Solanum mauritianum Solanum aviculare

100 N 2 1 90 5 80 115 70 1 60 Figure 2 Spatial distributions of seeds in seed 2 50 in soil seed bank within the one hectare 40 subtropical rainforest. The X Y coordinates are

30 the locations in the one hectare plot (m).

20 Numbers indicated number of seed germinated. The direction of North was indicated. 10 1 0 0 20406080100

Pseudoweinmannia lachnocarpa

Figure 6.2 Spatial distributions of seeds in the soil seed bank within the 1 ha subtropical rainforest plot. The X and Y coordinates are the locations in the 1 ha plot (m). Numbers indicate the number of seeds germinated from the sample at each location. The direction of North is indicated.

111

6.3.4 Distribution of seeds and canopy trees of Caldcluvia paniculosa

There were 66 canopy trees (DBH>5 cm) of C. paniculosa recorded in the entire hectare, of which 31 were considered of reproductive size (DBH>30cm). The distributions of these canopy trees differed across size classes (Table 6.6). If all trees are included in the analysis, the distribution shows a significant autocorrelation at 70 and 80 metres (Fig. 6.3), indicating that the northern half of the plot had more trees. Patterns for all trees are driven by large sized trees (DBH > 30 cm), which shows a strong aggregated distribution and significant autocorrelation at distances from 20 m to 40 m and 70 m to 90 metres. The 70 m to 90 m autocorrelation reflects a decrease in tree density from north to south and the 20 to 40 metre autocorrelation may be contributed by the two distinctive groups within the north half of the plot (Fig. 6.3). The trees below 30 cm DBH showed no autocorrelation at any distance (Table 6.6).

100

0 0 20406080100

15cm>DBH ≥5 cm 30 cm >DBH ≥15cm DBH ≥30 cm

Figure 6.3 Spatial distribution of the canopy trees (DBH > 5 cm) of Caldcluvia paniculosa within the 1 ha subtropical rainforest plot.

The soil samples with high seed densities of C. paniculosa were taken from plots where many large size (DBH > 30 cm) canopy trees were concentrated (Fig. 6.4). The

112

areas in the 1 ha plot, which lacked large size C. paniculosa also showed a lack of seed of C. paniculosa in the soil seed bank (Fig. 6.4). The average distance from sampling spots without C. paniculosa seed to the nearest conspecific large canopy tree was 17.1 ± 1.6 m (Mean ± SE, n = 19), which was significantly longer than that of 7.9 ± 0.61 m for sampling plots that had 5 or more seeds (n=39, student-T-test, P<0.001). There were five large-sized canopy trees (DBH>50cm), of species other than C.paniculosa, recorded in the initial survey in 1995 that had died before the resurvey in 2006 (Fig. 6.4). There was a large patch around the four dead canopy trees with many soil samples containing no C. paniculosa seeds (Fig. 6.4). Moreover, there was also no large size C. paniculosa tree found within this patch (Fig. 6.4).

100

0 0 102030405060708090100

Figure 6.4 Distribution patterns of seeds ( - 0 seed, ● 1-5 seeds, ● 5 or more seeds) in relation to large sized canopy trees of Caldcluvia paniculosa (Δ, DBH >30 cm) within the 1 ha subtropical rainforest plot. * shows the location of five large canopy trees (DBH> 50 cm, not C. paniculosa) that died during 1995 to 2006.The direction of North is indicated.

The densities of C. paniculosa seed were significantly negatively correlated with the distance from the sampling spot to the nearest large-size (> 30 cm DBH) conspecific canopy tree (Fig. 6.5, Spearman Rank Correlation, R=-0.401, P<0.001). The average distance to the first two large sized canopy trees showed an even stronger correlation (Spearman Rank Correlation, R=-0.538, P<0.01). The density of other seeds, excluding C. paniculosa, showed no correlation with distance to large sized trees of C.

113

paniculosa (Spearman rank correlation, R = 0.106, P > 0.293) nor to the nearest trees (all species) above 30 cm in DBH (Spearman rank correlation, R=-0.027, P=0.790).

5 Number of seeds germinated

Number of Seeds 0 0 5 10 15 20 25 30

Distance to nearest large canopy tree

Figure 6.5 Correlation of the seed density of Caldcluvia paniculosa in the soil seed bank with the distances to the nearest conspecific large canopy tree (DBH> 30 cm)

6.4 Discussion

2 The overall seed density in the soil seed bank (918 seed/ mP )P of this 1 ha plot was slightly higher than that found in nearby similar undisturbed rainforests (566 and 2 603/mP P in top 5 cm soil, Abdulhadi & Lamb 1988) which may be due to the high abundance of C. paniculosa seed recorded in this study. Mast seeding of C. paniculosa has been observed at time intervals of three to five years (Floyd 1990). This kind of mast fruiting which is induced by the periodical droughts in seasonal rainforest has been considered as an evolutionary strategy to satiate natural enemies, both seed predators and seedling herbivores (Janzen 1974; Ashton et al. 1988; Waller 1993; Williamson & Ickes 2002). One mast fruiting of C. paniculosa was recorded in 1966 at Border Range National Park after a prolonged drought from 1957 to 1966 (Floyd 1990). The present study area has been subject to prolonged drought from 2000, since when the annual rainfall has been below average. Moreover, only 707 mm rainfall (less than half of average annual rainfall) was recorded in 2002 which was the driest year ever recorded since rainfall monitoring began in 1916. It is likely that our

114

sampling in 2004 followed a mast fruiting of C. paniculosa induced by this prolonged drought.

2 A similar seed density (90 seeds / mP )P and richness (14) of exotic seeds were recorded both in this study and in a study in a nearby mature rainforest 20 years ago (Abdulhadi & Lamb 1988). The seed density of the exotic shrub (or small tree) wild 2 tobacco, S. mauritianum, was substantially higher in the present study (41 seed/ mP )P 2 than recorded in the earlier study, 0 and 4 seeds/ mP P , by Abdulhadi and Lamb(1988). This may reflect the heavy proliferation of this species around O’Reilly’s Guesthouse located within 2 km of the present study. The seeds are known to be distributed by various rainforest birds, such as the King Parrot and Satin Bowerbird (Church 1997), as well as flying fox (Pteropus spp).

Soil seed banks in primary rainforest are known to be dominated by short- lived pioneer species, which are unable to recruit beneath the mature forest canopy and may show a spatial or temporal separation of seeds and trees (Cheke et al. 1979; Putz 1983; Garwood 1989; Dalling et al. 1997). The soil seed bank in the present study, however, was predominantly composed of the long-lived pioneer tree species C. paniculosa, which ws a common canopy tree. Studies in adjacent rainforests have also found such species such as C. paniculosa and G. benthamii to be dominant, in both canopy and soil seed bank (Abdulhadi & Lamb 1988). These species are generally present as large canopy trees and can become locally dominant in this kind of subtropical rainforest at higher altitudes (Burgess et al. 1975; Floyd 1990; Smith et al. 2005). In a temperate rainforest of southern Chile, closely similar to Australian subtropical rainforest, a long-lived pioneer tree, Weinmannia trichosperma (Cunoniaceae) was dated at more than 750 years old (Lusk 1999). The importance of these long-lived pioneer species in terms of subtropical rainforest conservation and restoration needs further study. Short-lived pioneer species, such as R. rosifolius, S. aviculare, Duboisia myoporoides and Rhodomyrtus psidioides, are also common in soil seed bank but showed relatively low seed densities. These species may become dominant in early stages of secondary succession in subtropical rainforest in Southeast Queensland and northern New South Wales (Webb et al. 1972; Olsen & Lamb 1988; Floyd 1990).

Despite the seed distributions of some species (e.g. C. paniculosa) showing

115

significant correlations with the position of conspecific canopy trees, the structure of above-ground vegetation appears to have little effect on the local distribution of overall seed abundance. This suggests that other ecological processes other than local dispersal may be important in determining seed distribution in the soil seed bank. In a study of pioneer trees at Barro Colorado Island, Dalling et al (1998b) found that the local seed distribution was closely related to the distance to seed trees but, at a larger scale, this association seemed to have little impact on the spatial distribution of overall soil seed bank, due largely to the spatially or temporal separation of seeds and trees, causing by seed dormancy and gap dependent regeneration (Putz 1983; Swaine & Whitmore 1988; Garwood 1989). Many species found in the soil seed bank in present study, such as R. rosifolius and S. mauritianum, had no reproductive individuals in the plot. Their seeds, presumably, were transported there from a considerable distance and dropped, mainly by frugivores, or possibly are persistent seeds from seed plants that have died. Moreover, secondary dispersal, possibly by ants may also contribute to the distribution of seeds in the soil seed bank (Levey 1993).

In this study, the density of canopy lianas was negatively correlated with the density of non-woody and exotic seeds in the soil seed bank, while the density and total basal area of lianas were positively correlated with the species richness and basal area of trees (Chapter 4). This liana and tree association may have increased the complexity and coverage of the forest canopy, thereby reducing the likelihood of invasion by wind dispersed exotic seeds, from, for example, the family Asteraceae. The negative correlation between the tree species richness and the density of non-woody seeds also supports this hypothesis. These correlations may be also associated with forest gaps which play important roles in the dynamics of soil seed bank (Putz 1983; Dalling et al. 1998a).

The dominant canopy tree species, C. paniculosa, showed a strongly clumped distribution in this 1ha plot. There are several possible explanations for this distribution pattern. First, this distribution pattern may be caused by the special regeneration strategy of this pioneer species which requires forest gaps to germinate and grow (Burgess et al. 1975; Hopkins 1975; Floyd 1990; Lusk & Kelly 2003). The clumped distribution of C. paniculosa may therefore be a reflection of a past disturbance event, such as a cyclone that can cause considerable destruction and

116

promote the establishment of pioneer species (Olsen 1990).

Secondary, the distribution pattern may reflect the special regeneration niche of the small seeded C. paniculosa. Small seeds have been found to have the advantage of being able to use steep, litter-free habitats or even tree trunks as germination sites (Grubb & Metcalfe 1996). Some special habitats under intact rainforest, such as tree fern trunks and dead logs, can also provide seed germination and seedling growth needs for C. paniculosa (Floyd 1990). Lusk and Kelly (2003) found that the seedlings of C. paniculosa were strongly associated with elevated micro-sites in a Chilean temperate rainforest which may offset the disadvantage of their being too small to penetrate leaf litter. This special regeneration strategy may have also contributed to its distribution in this 1 ha plot as large sized trees were mainly confined to the southwest steep corner where many tree ferns Cyathea leichardtiana were also found.

Apart from the above two possible reasons, this study also found that seed dispersal limitation might have contributed to the clumped distribution of the canopy trees. Floyd (1990) classified C. paniculosa as wind dispersed due to its light weight (only0.1 - 0.2 mg), dust like seeds. The result of the present study suggests that this species may have limited dispersal ability. Seeds of C. paniculosa are dispersed around April to July (Floyd 1990). During peak fruiting season, a density of 50 seeds / 2 mP P was recorded in monthly seed rain near reproductive trees (Floyd 1990). The soil samples in the present study were taken in October. Many soil samples taken from where there were no large size C. paniculosa trees contained few or even no seeds whereas high seed densities were found from samples taken close to large size conspecific trees. In addition, the seed density decreased with the increasing distance from large size conspecific canopy trees. In a study of secondary subtropical rainforest, Stewart (1995) did not find any C. paniculosa seedlings germinating during the two years of seedling monitoring despite there being reproductive trees just outside her study site. Stewart (1995) ascribed this to a lack of special germination sites such as tree fern trunks. Seedlings of C. paniculosa, however, can always be found in tree fall gaps especially those with exposed mineral soils (Burgess et al. 1975; Hopkins 1975; Floyd 1990). It is likely that dispersal limitation caused the absence of C. paniculosa in Stewart’s (1995) secondary forest because the seed was absent from both the soil seed bank and the seed rain. As mentioned by Floyd (1990), many

117

canopy trees in Australian subtropical rainforest are wind dispersed (32%) including dominant species such as Argyrodendron spp and C. paniculosa; there species may have a limited dispersal distance except, perhaps, under unusually windy conditions.

6.5 Conservation applications

This study shows that the soil seed bank in this subtropical rainforest may have more influence on the forest dynamics than in many rainforests elsewhere. The soil seed bank contains not only short-lived pioneer species that can initiate secondary succession, but also long-lived pioneer species that could be potential large canopy trees. Rainforest soil could potentially be used in restoration projects to help restore rainforest canopy species. Apparently, a lack of seeds is often one of the biggest obstacles in rainforest restoration (Holl et al. 2000; Erskine 2002; Lamb et al. 2005). Those long-lived pioneer species should be included in restoration, to act as a seed source to promote forest regeneration.

118

7 SOIL SEED BANKS AT RAINFOREST EDGES

7.1 Introduction

It is widely accepted that rainforest soil seed banks are composed mainly of so-called pioneer species that require forest gaps to germinate (Cheke et al. 1979; Hopkins & Graham 1983; Whitmore 1983; Garwood 1989; Graham & Hopkins 1990). On the other hand, primary rainforest species generally have a fast germination strategy and rely more on seedlings persisting under dense canopy to regenerate rather than on seeds in soil seed banks (Vazquez-Yanes & Orozco-Segovia 1993; Garwood 1996; Whitmore 1996, 1998a). In primary rainforest, pioneer species generally contribute less than 3% of canopy trees (Whitmore 1989; Laurance et al. 2006b), but play important roles in rainforest regeneration, especially in rainforest recovery after disturbance (Bazzaz 1983; Whitmore 1983; Young et al. 1987; Olsen & Lamb 1988).

The seed density in soil seed banks of primary rainforest is maintained at a relatively low level with a high proportion of seeds from woody species (Hopkins & Graham 1983; Putz 1983; Garwood 1989; Graham & Hopkins 1990). Once the rainforests have been heavily disturbed, by factors such as cyclones, selective logging and rainforest clearing, the exposure of the soil seed bank to increased radiation and soil disturbance will lead to the germination and subsequent depletion of seeds in the original soil seed bank (Webb et al. 1972; Young et al. 1987; Olsen & Lamb 1988; Williams-Linera 1990a). This may in turn be followed by a dramatic increase in seed input, mainly from annual and exotic species from extrinsic sources (Hopkins & Graham 1984; Young et al. 1987; Abdulhadi 1989).

Rapid global rainforest deforestation is increasing the proportion of edges, exposing the rainforest interiors to influences from other landscapes that contrast in species composition and structure to rainforest (Turner & Corlett 1996; Whitmore 1997). Rainforest fragmentation is also changing the soil seed bank in a number of ways that, over the long term, may affect rainforest regeneration. First, changes in environmental conditions may have a profound impact on the prior existing soil seed bank. Elevated light levels and temperature, and increased soil disturbance may promote the

119

germination of seeds in the soil seed bank during the early stage of edge exposure (Young et al. 1987; Williams-Linera 1990a; Cubina & Aide 2001). In Amazonian rainforest, for example, dramatic increases in secondary tree species at edges may result from seed germinating from soil seed banks (Didham & Lawton 1999; Laurance et al. 2006b). In addition, changes in soil conditions (e.g. high temperature) near edges may produce conditions unsuitable for the persistence of some species in the soil seed bank. Second, large amounts of seed from surrounding matrices are continually brought into the rainforest, by means of wind or animal dispersal (Janzen 1983; Willson & Crome 1989; Cadenasso & Pickett 2001; Cubina & Aide 2001; Peters 2001). This may cause dramatic increases in seed density, especially of exotic species, in soil seed banks near rainforest edges (Willson & Crome 1989; Cubina & Aide 2001). The distance to which seeds may penetrate into the rainforest is largely determined by the structure of the edges (Cadenasso & Pickett 2001). Third, subsequent changes upon ecological processes after edge creation, such as seed dispersal and predation, may also cause substantial changes in the soil seed bank (Nepstad et al. 1996; Wijdeven & Kuzee 2000; Peters 2001; Terborgh et al. 2001). Currently, a decrease in abundance of seed dispersal agents, especially those that disperse large rainforest seeds, has become a threat to the regeneration of rainforest remnants (Terborgh & Nunez-Iturri 2006).

On the other hand, some rainforest species are dispersed into surrounding matrices and may become important seed sources promoting natural forest regeneration (Turner & Corlett 1996; Aide et al. 2000; White et al. 2004). A shortage of viable seeds has been identified as a primary constraint impeding the regeneration of deforested areas despite the close proximity to reliable seed sources (Holl 1998; Holl 1999; Wijdeven & Kuzee 2000; Cubina & Aide 2001). Contributing to this shortage of seeds may be the lack of suitable perch sites and food sources for dispersal agents in open pastures which eventually will limit seed dispersal (Willson & Crome 1989; Holl 1998; Toh et al. 1999). Even if seed dispersal occurs, rainforest seeds may not accumulate in the soil seed bank due to high predation pressures and harsh environmental conditions (Nepstad et al. 1996; Cuaron 2000; Cubina & Aide 2001).

The exchange of species between rainforests and surrounding vegetation can have important ecological consequences in terms of rainforest conservation as well as

120

restoration (Janzen 1983; Turner & Corlett 1996; Lamb et al. 2005). Understanding seed movement at rainforest edges requires study at the landscape scale because different surrounding matrices in a landscape can generate different patterns of seed rain and the edge effects that can influence seed movement tend to be edge specific (Murcia 1995; Harper et al. 2005).

This chapter compares the composition and distribution of seeds in soil seed banks across three different types of rainforest edges, within a fragmented subtropical rainforest complex. The main questions addressed were as follows:

1. Are the species composition and seed accumulation in soil seed banks different at different types of edges? 2. How does the composition and structure of above-ground vegetation affect the composition and seed distribution in the soil seed bank? 3. How are seeds distributed in the soil seed bank from edge to rainforest interior, and are the distribution patterns different across different rainforest edges? 4. What kinds of seeds have been dispersed from surrounding matrices into rainforest and how are the seeds distributed in the rainforest soil seed bank? 5. What kinds of seed have been dispersed from the rainforest into surrounding matrices and how are the seeds distributed in the soil seed bank of the surrounding matrices?

7.2 Methods

7.2.1 Data collection

Sampling of the soil seed bank was based upon nine transects established across each of the three types of rainforest edges (for sample methods see Chapter 2). Two soil samples (each 10 cm × 10 cm area and 5 cm deep), 10 m apart and perpendicular to the edge transect, were taken at each of five distances along the edge transect which stretched 100 metres into the rainforest. For each type of edge, a total of 90 soil samples were taken along the nine transects, with 18 soil samples from each of the five distances from edge to rainforest interior. In addition, 6 soil samples were also

121

taken in each of pasture and eucalypt forest matrices, where the edge transects extended 50 m from the rainforest edge (3 sampling distance on the transect, same arrangement as sampling in the rainforest). In total, 378 soil samples were taken with 90 samples from each of the three types of rainforest edges and 54 samples from eucalypt forests and pastures, respectively. The soil samples were then germinated to assess the seed composition and density in the soil seed bank in a shade house (germination methods are described in Chapter 2).

This chapter also investigates the relationship among soil seed banks and the above- ground vegetations and how this caries with edge type and distance from the edge into rainforest interiors. The data from vegetation surveys, including trees, lianas and seedlings, were used in the analysis and the datasets are summarized in Table 2.1.

7.2.2 Data analysis

Seed density in this study is defined as the total number of seeds present in one square metre of soil to a depth of 5 cm. The species richness in soil seed banks was estimated using a first-order Jack Knife estimate using the program EstimateS (Colwell 2005). Shannon diversity indices (Magurran 2004) were also calculated using the total germinated seedlings from the edges and the surrounding matrices.

The distribution of seed abundance from edge to rainforest interior and among edges was compared using two-way analysis of variance (ANOVA), treating edge type and distance from edge to rainforest interiors as independent factors. In addition, the distance effect on seed distribution was further assessed using Spearman Rank Correlation based on numbers of seedlings germinated.

The relationship between the abundance and number of species in the soil seed bank and above ground vegetation in the associated survey plots were assessed using Spearman Rank correlation.

The comparison of species composition among soil seed banks at different types of edges was performed using a non-metric multi-dimensional scaling ordination (NMDS) based on Bray-Curtis similarity measure (Clarke & Warwick 2001).

122

Analyses of similarity (ANOSIM) were also carried out to assess the between edge differences. In addition, the associations of each of the species with the ordination coordinates was assessed using multiple regression (Pop Tools Excel add-in, Hood, 2003), and the significance levels of the regression were further evaluated suing a randomization test (Edgington 1980). For the species that showed significant associations, a biplot vector was then generated in ordination space with x and y coordinates from Pearson’s correlation indices between the seed abundance and the scores of ordination axes.

7.3 Results

7.3.1 Summary of germination experiments

A total of 9299 seeds were germinated from all soil samples. Of this total, 5857 seeds were from soil samples taken from active pastures and eucalypt forests near the rainforest edges (n = 108), and 3442 seeds from soil samples within the rainforest edges (n = 270). The mean number of seedlings germinated from each of the soil samples was 24.6 ± 2.6 (Mean ± SE). One hundred and forty five species from 54 families were identified, of which 103 species were native and the other 42 species considered to be exotic, the majority of them being agricultural weeds (Table 7.1). Exotic seeds made up more than 65 percent of the germinated seedlings, mainly non- woody species associated with grazing (e.g. Oxalis corniculata, Cyperus brevifolius) and many species from the family Asteraceae. Native tree species made up 89% of the total tree seeds germinated. The other 11% of tree seeds were from the small exotic tree species Solanum mauritianum.

Poaceae was the most species rich family in the soil seed bank with 18 species, followed by Asteraceae and Cyperaceae with 16 and 11 species respectively. When the number of successfully germinated seeds alone is considered, Asteraceae dominates, making up 23 percent of all seeds germinated. Together, the families Asteraceae, Oxalidaceae and Cyperaceae, make up 58 percent of total seed germinated. Seedlings from these three families were mainly exotic (95%) and accounted for 84 percent of the exotic seeds.

123

The most dominant species were O. corniculata and C. brevifolius. Both of the two exotic species are known as weed species frequently associated with pasture (Harden, 1993). Combined, they made up 34 percent of total germinants and 53 percent of total exotic seeds in the soil seed bank. The dominant tree species was Dendrocnide excelsa which accounted for 42 percent of all tree seeds germinated. The shrubs were dominated by Rubus rosifolius and a few species from genus Solanum, such as the kangaroo apple Solanum aviculare.

Table 7.1 Summary of germination experiments with dominant plant families according to their contributions to species richness or seed density in the soil seed banks, from soil samples taken from transects near the rainforest edge (n = 270) and from pature and eucalypt forest ( n = 108). Each sample was 10 cm ×10 cm area, 5 cm deep. Herb Grass Tree ShrubVine Native Exotic No. of Species Poaceae 0 18 0 0 0 10 8 Asteraceae 16 0 0 0 0 6 10 Rutaceae 0 0 1 10 0 11 0 Cyperaceae 0 9 0 0 0 6 3 Solanaceae 2 0 4 2 0 4 4 Apiaceae 5 0 0 0 0 3 2 Moraceae 0 0 0 3 2 5 0 Rosaceae 1 0 1 0 3 3 2 Rubiaceae 1 0 1 1 1 4 0 Amaranthaceae 2 0 1 0 0 2 1 Other 44 families 22 1 5 16 17 49 12 Total 49 28 13 32 23 103 42 No. of germinates Asteraceae 2095 0 0 0 0 214 1881 Oxalidaceae 1677 0 0 0 0 0 1677 Cyperaceae 0 1591 0 0 0 65 1526 Poaceae 0 582 0 0 0 336 246 Solanaceae 158 0 219 131 0 228 280 Apiaceae 496 0 0 0 0 483 13 Urticaceae 46 0 0 435 0 435 46 Rosaceae 36 0 312 0 32 343 37 Juncaceae 0 337 0 0 0 337 0 Cunoniaceae 0 0 0 302 0 302 0 Other 44 families 525 0 118 168 39 476 374 Total 5033 2510 649 1036 71 3219 6080

Twenty-seven species of vine were germinated which made up 19% of the species richness but less than 1% of seeds in the soil seed bank. Most of the vine species had only a few seedlings germinated in the experiment, except one species, Rubus moorei,

124

which was found across all three rainforest edges with 29 seeds germinated in total. The only exotic species of vine, Rubus ellipticus (McDonald & Thomas 1990), had just a single seed germinated.

7.3.2 Composition of soil seed banks

Soil seed bank within rainforest edges The soil seed banks in rainforest at the pasture/ rainforest edges had the highest seed density, more than three times than that at the the eucalypt forest/ rainforest edges and twice that at the roadside edges (Table 7.2).

Table 7.2 Compositions of the soil seed banks of rainforest edges (Seed densities are Mean ± SE seeds per square metre of top5cm of soil; different subscript letters in the same line indicate a significant difference). (Based on the germination of 90 soil samples of 10 cm × 10 cm in area for each edge types and 54 soil samples for eucalypt forest and pasture respectively) , Rainforest edges Non-rainforest Roadside/ Pasture/ Eucalypt Eucalypt Pasture rainforest rainforest forest/ forest rainforest Number of samples 90 90 90 54 54 a b c Seed density 1005 ± 83P P 2127 ± 24P P 691 ± 72P P 712 ± 81 10135±882 a b c Non-woody seed density 320 ± 36P P 1533 ± 248P P 196 ± 26P P 467 ± 76 10107 ± 884 a a b Woody seed density 686 ± 78P P 593 ± 46P P 496 ± 66P P 244 ± 39 28 ± 19 a a b Native seed density 812 ± 80P P 946 ± 142P P 537 ± 64P P 613 ± 79 1522 ± 232 a b a Exotic seed density 193 ± 24P P 1181 ± 154 P P 153 ± 23P P 98 ± 19 8613 ± 786 a ab b Tree seed density 505 ± 76P P 297 ± 35P P 283 ± 38P P 89 ± 21 19 ± 19 No. of native species 46 52 50 21 34 No. of exotic species 18 31 14 14 33 Species diversity 2.89 3.19 3.1 2.72 2.34 Observed species richness 64 83 64 54 48 Estimated species richness 88.7 106.7 88.7 65.8 69.6

The high seed accumulation at the pasture/ rainforest edges, however, was primarily composed of non-woody seeds with only 38 % seeds from woody species. In contrast, the soil seed bank at the roadside edge and eucalypt forest/ rainforest edge had high proportions of seed from woody species, making up 68 and 72 % of total seeds, respectively. The soil seed bank at roadside edges had a similar density of woody seeds as at the pasture/ rainforest edges, that is, slightly higher than that at the eucalypt forest/ rainforest edge.

125

The soil seed banks at the roadside and eucalypt forest/ rainforest edge were dominated by seeds of native species, contributing more than 78% of the seeds germinated. In contrast, at the pasture/ rainforest edges, exotic seeds dominated the soil seed bank with more than 56% of germinated seeds from species exotic to the area. Some of them were notorious weeds, such as Ageratina riparia and Senecio madagascariensis. The density of exotic seeds was much higher at pasture/ rainforest edges than at the roadside/ rainforest and eucalypt forest/ rainforest edges.

Pasture/ rainforest edges appeared to hold the most species rich soil seed bank. However, nearly one third of all seeds were from exotic species, especially species from the families Poaceae, Asteraceae and Cyperaceae that are closely associated with cattle farming. Nearly the same numbers of native species were found across the three types of edges. The eucalypt forest/ rainforest and pasture/ rainforest edges showed slightly higher species diversity than the roadside edge.

Soil seed bank in the surrounding matrices The surrounding matrices, i.e., pasture and eucalypt forest, differed substantially in the composition of their soil seed banks when compared with the adjacent rainforests (Table 7.2).

Table 7.3 Numbers of tree seedlings germinated from the soil seed banks from eucalypt forests and pastures adjacent to rainforest edges. (Germination of 54 10 cm × 10 cm soil samples of top 5 cm soil for each of the two site types) Species Family Eucalypt forest Pasture Dendrocnide excelsa Urticaceae 28 0 Solanum mauritianum Solanaceae 4 10 Polyscias elegans Araliaceae 6 0 Pseudoweinmannia lachnocarpa Cunoniaceae 4 0 Trema tomentosa Ulmaceae 2 0 Acacia melanoxylon Fabaceae 2 0 Homalanthus nutans Euphobiaceae 1 0 Acronychia oblongifolia Rutaceae 1 0

The pasture had a soil seed bank nearly four times higher in seed density than the nearby rainforest edge. Those seeds were mainly from exotic species and these made up 85% of the germinated seeds. No native rainforest tree seed was found in the soil sample from the pastures. Only one exotic tree species, wild tobacco (S. mauritianum),

126

was germinated from pasture soil, and from only a single soil sample. Most of the exotic species found in the pasture were also recorded in the soil seed bank of nearby rainforest edges.

The seed density in the soil seed bank of eucalypt forest showed no significant difference from that of the nearby rainforest edge (One-way ANOVA, P = 0.364) but the species composition of the soil seed banks differed significantly. The seeds germinated from eucalypt forest were mainly non-woody species, primarily grasses (Poaceae) which dominated the understorey of the eucalypt forest. The density of tree seeds was significantly lower than in the rainforest (One-way ANOVA, P<0.001). There were only eight species of trees that germinated from the soil samples from eucalypt forest, most of which had only a few seeds except for the giant stinging tree (Dendrocnide excelsa) with 28 seeds recorded (Table 7.3). Most of those species are also present as seedlings or saplings under the eucalypt forest understorey near the rainforest edges.

Both the pastures and eucalypt forest showed a lower species diversity in their soil seed banks than in the nearby rainforests (Table 7.2).

7.3.3 Species estimation

The species estimation procedure suggested a figure of approximately 200 species in the soil seed bank of this rainforest dominated landscape when the samples from rainforest and the surrounding pasture and eucalypt forest were all combined in the estimation.

The pasture/ rainforest edge showed a higher species accumulation than the other two types of edges (Fig. 7.1). The estimation curves of eucalypt forest/ rainforest and roadside edge were similar. Surprisingly, the eucalypt forest showed a similar curve to that of the nearby rainforest. The curve for the soil seed bank of the pasture was apparently reaching an asymptote, suggesting that the sample size may have been enough to capture most of the species. None of the other curves showed a tendency to approach an asymptote, indicating that more species could be expected if a larger sampling effort was considered.

127

120

100

Pasture/rainforestPasture/ rainforest edge edge 40 RoadsideRoadside/ edge rainforest edge EucalyptusEucalypt forest/ forest/rainforest rainforest edge edge Eucalyptus forest Cumulative number of species of species number Cumulative 20 Eucalypt forest PasturePasture

0 020406080100

Number of sampling plot Figure 7.1 Species accumulation curves based on first-order jackknife estimations on the seedlings germinated from the soil seed banks of rainforest edges and the surrounding matrices. (90 samples for each of edge type and 54 samples for the eucalypt forest and the pasture, respectively)

7.3.4 Seed distribution along edges

The densities of seeds in the soil seed banks were significantly different among edges and showed a strong distance effect without an interaction between edge type and distance effect (Table 7.4).

Table 7.4 Comparisons of the number of seedlings germinated from the soil seed banks at different positions from edge to rainforest interiors at different types of rainforest edges. (18 samples for each of the five distances, Two-way ANOVA, significant results are highlighted) Edge type Distance Edge type * Distance df = 2 df = 4 df.= 8 F-ratio P-value F-ratio P-value F-ratio P-value All seeds 38.465 <0.001 3.959 0.004 1.846 0.069 Exotic seeds 93.936 <0.001 3.411 0.010 3.942 <0.001 Native seeds 11.088 <0.001 3.008 0.019 0.982 0.451 Tree seeds 4.061 0.018 4.632 0.001 1.737 0.090

The pasture/ rainforest edge showed consistently higher seed accumulation than the other two types of edges up to 60 metres into the rainforest (Fig. 7.2). The highest seed density was found at about 5 m into the pasture. The seed density at the pasture/ rainforest edge was correlated negatively with the distances from pasture to rainforest

128

interiors (Pearson correlation, R=-0.705, P<0.001). In contrast, the seed densities at eucalypt forest/ rainforest and the roadside edges showed no correlation with distance (Pearson correlation, R =0.004, P=0.967, R= -0.139, P= 0.193, respectively)

Seed density (/m2)

16000

Eucalypt forest/rainforest edge 12000 Pasture/rainforest edge Roadside edge

8000

4000

0 -50 -25 0 25 50 75 100

Non-rainforest Distance (m) Rainforest

Figure 7.2 Distributions of seeds in the soil seed banks near three types of rainforest edges. (Each data point represents the average seedling germinated from 18 soil samples, error bars are standard errors)

The distribution pattern of exotic seeds showed a similar pattern to that of total seeds (Fig. 7.3). A strong interaction between edge type and distance effect suggests that the distributions of exotic seeds were different at different types of edges (Table 7.4).

The density of exotic seeds showed a sharp drop from pasture to the rainforest within a few metres (Fig. 7.3) and the densities were negatively correlated with the distances from edge to rainforest interior (Fig. 7.3, Spearman Rank Correlation, R= -0.415, P<0.001). In contrast, the densities of exotic seeds were similar at different distances across the eucalypt forest/ rainforest and roadside edges and also showed no difference between eucalypt forest and the nearby rainforest.

The density of native seeds showed a significant difference among edge types and a distance effect without an interaction (Table 7.4). All the edges showed a slightly higher density at the first distance into the rainforest and showed a similar density at other distances within the rainforest (Fig. 7.4). The soil seed bank in the pastures showed a higher density of native seed than that within the rainforest except that at

129

the first distance (Fig. 7. 4). No significant difference was found between eucalypt forest and nearby rainforest in terms of the density of native seed (Fig. 7.4).

Seed density (m2) 12000

10000 Eucalypt forest/rainforest edge Pasture/rainforest edge 8000 Roadside edge

6000

4000

2000

0 -50 -25 0 25 50 75 100

None-rainforest Rainforest Distance (m)

Figure 7.3 Distributions of exotic seeds in the soil seed banks near rainforest edges. (Each data point represents the average seedling germinated from 18 soil samples, error bars are standard errors)

The densities of tree seeds showed a significant difference across edge types and a distance effect without interaction between the factors (Table 7.4, Fig. 7.5). The roadside edges showed a higher density than the eucalypt forest/ rainforest edge whereas the pasture/ rainforest edge showed no significant difference from both of them (ANOVA, P=0.018, Tukey’s HSD). The density of tree seeds showed a significant decrease from rainforest interior to the edge though the surrounding matrices at the pasture/ rainforest and eucalypt forest/ rainforest edges. This decrease was significantly correlated with distance from rainforest interior into the surrounding matrices at both edges (Pearson correlation, R=0.653 and 0.444, respectively, both P<0.001). A few samples at the first two distances into rainforest at roadside edges contained high densities of seeds of D. excelsa. This caused a large variation in seed densities (Fig. 7.5). If seeds of D. excelsa are removed from analyses, the density of other seeds is then correlated significantly with the distance from the rainforest interior to the edge, showing the same pattern as found at other two type of edges (Pearson correlation, R = 0.264, P=0.012).

130

Seed density (m2)

3000

2500 Eucalypt forest/rainforest edge Pasture/rainforest edge 2000 Roadside edge

1500

1000

500

0 -50 -25 0 25 50 75 100

None-rainforest Rainforest Distance (m)

Figure 7.4 Distributions of native seeds in the soil seed banks near rainforest edges. (Each data point represents the average seedling germinated from 18 soil samples, error bars are standard errors)

Seed density (m2)

900 Eucalypt forest/rainforest edge Pasture/rainforest edge 750 Roadside edge

600

450

300

150

0 -50 -25 0 25 50 75 100

None-rainforest Rainforest Distance (m)

Figure 7.5 Distributions of tree seeds in the soil seed banks near rainforest edges. (Each data point represents the average seedling germinated from 18 soil samples, error bars are standard errors)

131

7.3.5 Response of individual species to distance from the edge

Correlations of seed density with the distance from the edge to the rainforest interior demonstrated different distribution patterns for different species (Table 7.5).

Table 7.5 Correlations of the densities of dominant species in the soil seed banks with the distances from edges to the rainforest interiors. (270 samples from three types of rainforest edges, 90 samples for each of the five distances) Species Germinates Pearson's R P Ageratina riparia 43 -0.161 0.008 Caldcluvia paniculosa 185 0.187 0.002 Chenopodium carinatum 55 -0.040 0.512 Clematis glycinoides 42 -0.092 0.134 Conyza canadensis 428 -0.223 0.000 Commelina cyanea 43 -0.272 0.000 Cyperus brevifolius 181 -0.176 0.004 Cyperus tetraphyllus 25 -0.175 0.004 Dendrocnide excelsa 407 -0.036 0.557 Duchesnea indica 31 -0.046 0.455 Ficus watkinsiana 34 -0.038 0.537 Gamochaeta spicata 74 -0.160 0.008 Hydrocotyle pedicellosa 82 -0.164 0.007 Juncus flavidus 217 -0.145 0.017 Nyssanthes diffusa 30 -0.096 0.117 Oxalis corniculata 60 -0.062 0.314 Phytolacca octandra 54 -0.113 0.063 Polyscias elegans 39 0.086 0.156 Pseudoweinmannia lachnocarpa 113 0.212 0.000 Rubus moorei 26 -0.116 0.056 Rubus rosifolius 282 0.126 0.038 Senecio bipinnatisectus 76 -0.197 0.001 Solanum aviculare 167 -0.047 0.446 Solanum mauritianum 102 0.090 0.142 Solanum americanum 117 -0.068 0.268 Solanum nigrum 29 0.007 0.904 Solanum stelligerum 21 -0.082 0.178 Urtica incisa 45 0.147 0.016

Species that showed a significant positive correlation with the distances from the edge to the rainforest interior were, in general, woody species. These include two wind- dispersed tree species, Caldcluvia paniculosa and Pseudoweinmannia lachnocarpa, and a bird dispersed shrub, R. rosifolius. The species with decreasing seed abundance from edge to rainforest interior were mostly non-woody and, predominantly exotic species. Most of these species were dispersed from surrounding matrices into rainforest, such as many species from the Asteraceae and Cyperaceae. The distribution of most frugivore dispersed species, such as D. excelsa, Solanum spp. and Polyscias elegans, did not show any relationship, either negative or positive, with respect to the edge. An exotic herb, Urtica incisa, also showed an increase in seed density from the

132

edge to the rainforest interior.

2 Eucalypt forest/rainforest edge Stress=0.15 1.5 Pasture/rainforest edge

Roadside edge 1

0.5

-0.5

-1

-1.5

-2 -1.5 -1 -0.5 0 0.5 1 1.5

1 2 1 3 4 5 6 8 7 9 10

Figure 7.6 MDS analysis of rainforest edges based on the overall seedlings germinated from the soil seed banks (above). The vectors indicate increases in the density of species showing significant associations with the ordination axes (Below). Species showed significant association (p <0.05) with the ordination coordinates were: 1. Ageratina riparia 2. Rubus rosifolius 3. Senecio bipinnatisectus 4. Phytolacca octandra 5. Duchesnea indica 6. Conyza canadensis 7. Gamochaeta spicata 8. Juncus flavidus 9. Cyperus brevifolius 10. Oxalis corniculata

7.3.6 Comparison of soil seed bank composition among edges

The results of an MDS clearly separated the study transects into three groups, according to the edge types (Fig. 7.6). ANOSIM confirmed the observed patterns (Table 7.6). There were 24 species that showed significant associations with the ordination coordinates which were all associated with the pasture/ rainforest edges (Fig. 7.6). These species are mostly weed species growing in the pasture as well as

133

some pioneer rainforest species such as Phytolacca octandra and R. rosiflius. None of the tree species showed significant correlation with the ordination coordinates suggesting that the differences in the soil seed banks were mainly attributable to those short-lived species from surrounding matrices.

Repeating the same analysis with only tree seeds demonstrated again an obvious separation of roadside edges and eucalypt/ rainforest edges, but not pasture/ rainforest edges (Fig. 7.7). ANOSIM results also confirmed the ordination pattern (globe R=0.607, P = 0.001, Table 7.6). Eight tree species showed significant associations with the ordination coordinates which were found exclusively at roadside edges except for C. paniculosa and Polyscias murrayi which were also found in pasture/ rainforest edge (Fig. 7.7). Other dominant species, such as D. excelsa, P. lachnocarpa and P. elegans were common across all the edges and showed no significant association with any one of the three edges.

Table 7.6 Results of the ANOSIM on the composition of soil seed banks among different type of rainforest edges (Global R=0.607, P=0.001) All species Tree species Source* R- statistic P R- statistic P Euca.VS Past. 0.639 0.001 0.053 0.225 Euca.VS Road. 0.509 0.002 0.318 0.002 Past.VS Road. 0.679 0.001 0.107 0.112 * Euca. : Eucalypt forest/ Rainforest edge; Past.: Pasture/ Rainforest edge; Road. : Roadside rainforest edge

7.3.7 Soil seed bank and the rainforest canopy

The structural characteristics and species composition of the rainforest canopy, including trees and lianas, showed poor correlation with the seed density and species composition of the soil seed bank (Table 7.7). None of the variables measured in the rainforest canopy showed consistent correlation with the soil seed banks. Some of the variables, however, appeared to be good indicators of certain characteristics of the soil seed bank at certain types of edges.

At pasture/ rainforest edges, the number of tree species and tree diversity indices were negatively correlated with the density of woody seeds in the soil seed bank (Table 7.7). Conversely, the number of tree species was significantly but positively correlated with

134

the number of species in the soil seed bank at roadside edges. There was no significant correlation between the soil seed bank and the forest canopy at eucalypt forest/ rainforest edges.

2 Eucalypt forest/rainforest edge Stress=0.19

1.5 Pasture/rainforest edge

Roadside edge 1

0.5

-0.5

-1

-1.5 -1.5 -1 -0.5 0 0.5 1 1.5 2

3 65 7 4 8

Figure 7.7 MDS analysis of rainforest edges based on the tree seedlings germinated from the soil seed banks (above). The vectors indicate an increase in density of species that showed significant associations with the ordination axes (Below). Tree species were: 1. Polyscias murrayi 2. Quintinia verdonii 3. Acronychia octandra 4. Caldcluvia paniculosa 5. Rutaceae sp 6. Acronychia laevis7. Atractocarpus benthamianus 8. Acronychia suberosa

Eighteen out of 21 tree species in the soil seed banks had at least one stem recorded in the tree canopy (DBH>5cm) across the three rainforest edges (Table 7.8 ). There were three species, D. excelsa, P. elegans and P. lachnocarpa, which showed large seed accumulations in the soil seed bank and also a reliable canopy occurrence across the three edges. Another dominant species in the soil seed banks of the pasture/ rainforest and roadside edges, C. paniculosa, did not occur in either the soil seed bank or tree canopy at eucalypt forest/ rainforest edges.

135

Tree species shared by tree canopies and soil seed banks represented 84.5% of total tree seeds in soil seed banks of the three rainforest edges. In contrast, they contributed 18.5% of total stems and 37.4% of total basal area among the canopy trees (DBH > 5 cm) but only 5.8% of total stems of small trees (DBH 1 to 5 cm). Most of the species recorded in the tree survey of rainforest edges, 145 species of the 166 species, were not recorded in the samples of the soil seed bank.

Table 7.7 Spearman Rank Correlations between the seed density and number of species in the soil seed banks with variables from above-ground vegetation (45 samples for each of the edge types, data of two soil samples collected at each distance were pooled, significant results are highlighted, * P<0.05, ** P<0.01). None-woody Total Seed Species Woody seed Native seed Exotic seed seed Eucalypt forest/ rainforest edge Number of tree Tree species 0.154 0.097 0.188 -0.080 0.228 -0.061 Total tree basal area 0.017 -0.047 0.056 -0.169 0.004 -0.041 Tree diversity 0.171 0.137 0.241 -0.120 0.292 -0.113 No. of canopy lianas -0.026 -0.046 0.023 -0.148 0.034 -0.135 Species of seedling -0.203 -0.178 -0.175 -0.061 -0.134 -0.168 No. of liana seedling 0.190 0.154 0.260 -0.078 0.214 0.114 No. of tree seedling -0.372* -0.410** -0.138 -0.358* -0.320* -0.238 Pasture/ rainforest edge Tree species 0.129 0.124 -0.370* 0.156 0.243 0.024 Total tree basal area 0.128 0.127 -0.058 0.090 0.149 0.118 Tree diversity 0.090 0.056 -0.335* 0.111 0.207 -0.003 No. of lianas 0.242 0.060 -0.232 0.251 0.232 0.204 Species of seedling -0.417** -0.268 0.099 -0.496** -0.298* -0.443** No. of liana seedling -0.137 -0.042 -0.044 -0.142 -0.074 -0.147 No. of tree seedling -0.180 0.051 -0.141 -0.053 -0.187 -0.141 Roadside edge Tree species 0.267 0.304* 0.278 0.024 0.246 0.129 Total tree basal area 0.090 0.161 0.184 -0.081 0.124 0.037 Tree diversity 0.230 0.243 0.278 0.005 0.239 0.060 No. of lianas 0.036 -0.040 -0.162 0.191 0.019 -0.116 Species of seedling 0.292 0.251 0.232 0.297* 0.225 0.293 No. of liana seedling -0.133 -0.157 -0.281 0.180 -0.204 0.201 No. of tree seedling 0.399** 0.255 0.285 0.335* 0.340* 0.215

Nine to 11 species were shared by the soil seed bank and the canopy trees (DBH>5 cm) at each of the three types of rainforest edges (Table 7.9). These species composed nearly 80% of total tree seeds in the soil seed banks and were also frequently found in the tree canopy, contributing around 30% percent of total tree basal area. This high total basal area contribution was mainly made by some large, long-lived pioneer tree species such as C. paniculosa and D. excelsa. The pasture/ rainforest edge apparently had a larger proportion of tree stems and total tree basal area of those species shared by the soil seed bank and trees in above-ground vegetation than other edge types. The

136

low values of Sorensen’s indices suggest that the tree species compositions of the soil seed banks differed substantially from that of the above ground vegetation.

Table 7.8 Number of canopy trees (DBH>5cm) and number of seeds of species found in both the soil seed bank and the tree canopies. (45 10 m× 10 m tree plots and 90 soil samples for soil seed bank of each types of edges) Eucalypt forest/ rainforest Pasture/ rainforest Roadside Tree Seed Tree Seed Tree Seed Acronychia laevis 9 0 8 1 4 4 Acronychia pauciflora 2 1 0 0 0 0 Acronychia octandra 0 0 0 0 4 3 Acronychia pubscens 0 0 2 2 6 0 Acronychia suberosa 4 0 10 0 8 7 Alphitonia excelsa 2 2 15 0 1 0 Atractocarpus benthamianus 2 0 7 0 22 5 Caldcluvia paniculosa 0 0 15 39 5 146 Dendrocnide excelsa 6 111 20 94 7 202 Melicope micrococca 1 1 32 7 0 0 Ficus watkinsiana 1 12 1 18 1 4 Gejera salicifolia 3 1 4 1 1 0 Polyscias elegans 28 17 34 14 2 8 Polyscias murrayi 1 1 0 0 0 1 Vitex lignum-vitae 13 0 11 0 5 2 Pseudoweinmannia lachnocarpa 46 75 38 25 21 13 Quintinia verdonii 0 0 4 2 1 2 Zanthoxylum brachyacanthum 0 0 4 1 5 0

The proportions of stems of those species shared by the soil seed bank and the trees in above-ground vegetation increased following the tree size class at all three edges (Fig. 7.8). When compared among edges, this proportion was higher at the roadside edge for trees <10 cm. Notably, both the eucalypt forest/ rainforest and pasture/ rainforest edges showed a much higher proportion of those shared species than the roadside edge in medium-sized trees (DBH 20 to 30 cm), indicative vigorous secondary succession at those edges. Moreover, the pasture edge showed a consistently higher proportion of those shared species than the other two edges among tree assemblages ≥ 10 cm DBH.

7.3.8 Soil seed bank and seedling bank

The soil seed bank showed significant but negative correlation with the species richness or seedling density in the seedling bank at pasture/ rainforest and eucalypt/ rainforest edges. In contrast, the correlation between the seed and seedling was positive at roadside edges (Table 7.7).

137

At eucalypt forest/ rainforest edges, only the densities of the seedlings were significantly but negatively correlated with the number of species and seed density in the soil seed bank (Table 7.7). Seedling density was correlated in a like manner with the densities of non-woody seeds and native seeds in the soil seed bank.

Table 7.9 Summary of tree species shared by the soil seed bank and tree canopy at three types of rainforest edges. Eucalypt forest/ rainforest Pasture/ Roadside Site rainforest No. of species 9 11 11 Total seeds recorded 221 204 396 % of total tree seeds 0.867 0.761 0.853 % total canopy trees 0.124 0.177 0.134 % of total basal area 0.319 0.334 0.290 Sorensen’s s 17.5 18.3 19.8

At pasture/ rainforest edges, only the species richness of the seedling bank was negatively correlated with the seed density in the soil seed bank including the seed density of all, non-woody, native and exotic seeds (Table 7.7).

0.4

0.35 Eucalyptus forest/rainforest edge

0.3 Pasture/rainforest edge

0.25 Roadside edge 0.2

0.15

0.1

Percentage Percentage 0.05

0 1-5cm 5-10cm 10-20cm 20-30cm 30 cm and Tree size classes in DBH above

Figure 7.8 Proportions of trees in different size-classes of species shared by the soil seed banks and the above-ground vegetation at the rainforest edges.

At the roadside edge, the number of species in the seedling bank was positively correlated with the non-woody seeds and also correlated marginally with the density of total seeds and exotic seeds. Meanwhile, the density of tree seedlings showed a

138

positive correlation with the density of total seeds, non-woody seeds and native seeds.

There were 19 out of 32 tree species in the soil seed bank that had at least one individual represented as seedlings across the three types of edges (Table 7.10). These species accounted for more than 90% of total tree seeds germinated from soil samples.

Table 7.10 Occurrence of tree species found in both the soil seed and the seedling 2 banks at three types of rainforest edges (45 pooled 5 mP P seedling plots and 90 10 cm × 10 cm of top 5 cm soil samples for each of the edge types) Eucalypt forest/ rainforest Pasture/ rainforest Roadside Seedling Seed Seedling Seed Seedling Seed Acronychia laevis 1 0 5 1 4 4 Acacia melanoxylon 0 2 0 3 0 1 Acronychia oblongifolia 3 1 0 0 1 0 Acronychia octandra 0 0 0 0 2 3 Acronychia pauciflora 3 1 0 0 0 0 Acronychia pubscens 0 0 0 2 4 0 Acronychia suberosa 2 0 3 0 4 7 Alphitonia excelsa 1 2 0 0 0 0 Aphananthe philippinensis 0 0 1 1 2 0 Atractocarpus benthamianus 0 0 12 0 21 5 Caldcluvia paniculosa 0 0 0 39 0 146 Dendrocnide excelsa 1 111 0 94 2 202 Duboisia myoporoides 0 0 0 2 0 13 Elaeocarpus obovatus 1 0 0 1 0 1 Melicope micrococca 0 1 3 7 1 0 Ficus coronata 0 1 0 1 0 0 Ficus superba var. henneana 1 0 0 0 0 3 Ficus watkinsiana 1 12 0 18 0 4 Gejera salicifolia 0 1 2 1 0 0 Lophostemon confertus 0 4 0 0 0 5 Homalanthus nutans 2 2 1 1 0 1 Petalostigma pubscens 0 0 0 0 0 1 Plectranthus argentatus 0 0 0 0 0 9 Polyscias elegans 4 17 4 14 4 8 Polyscias murrayi 0 1 0 0 0 1 Premna lignum-vitae 1 0 0 0 2 2 Pseudoweinmannia lachnocarpa 44 75 79 25 2 13 Quintinia verdonii 0 0 1 2 4 2 Rubus rosifolius 0 0 0 0 0 1 Solanum mauritianum 0 17 0 53 0 32 Trema tomentosa 1 6 0 2 0 0 Zanthoxylum brachyacanthum 0 0 1 1 0 0

Most of the species in the seedling bank, 141 species, had no seed germinating from the soil samples. On the other hand, only 3 seedlings of D. excelsa were found during the plant survey despite the observation that there were hundreds of seeds germinated from soil samples. Likewise, C. paniculosa had many seeds in soil seed banks of pasture/ rainforest and roadside edges, but not a single seedling was found in the seedling bank. In addition, many short lived pioneer species such as Duboisia

139

myoporoides and Homalanthus nutans, did not occur in the seedling bank. Many of these species can be found at early stages in large rainforest gaps and in open areas outside the rainforest and are always present as early arrivals after large areas of rainforest disturbance, such as a cyclone (Webb et al. 1972; Olsen 1990). One of the large canopy tree species, P. lachnocarpa, had many seeds in the soil seed bank and many seedlings in the seedling bank as well. Most of these seedlings of P. lachnocarpa were recently germinated and had only a few pairs of leaves.

Table 7.11 Summary of species found in both the soil seed and the seedling banks at rainforest edges. Site Eucalypt forest/ rainforest Pasture/ rainforest Roadside No. of species 9 9 9 % of total tree seed 0.965 0.966 0.924 % total seedling 0.127 0.208 0.08 Sorensen’s S 17.8 19.1 18.2

For each type of edge, however, only nine species were shared by the soil seed bank and the seedling bank, reflecting different germination or establishment requirements among these tree species (Table 7.10, Table 7.11). Over 20% of seedlings recorded at pasture/ rainforest edges were of those species also found in the soil seed bank whereas the comparable figure was only 8% at the roadside edge. The values of Sorensen’s similarity indices were generally low across all three types of edges showing the dissimilarity between the seedling bank and soil seed bank.

7.4 Discussion

The results from this study show that species composition and seed density differed significantly among different type of rainforest edge. The composition and structure of the edges can have a strong impact in determining seed distribution in the soil seed bank. This has led to the changes in the species composition and seed accumulation in the soil seed banks from rainforest interiors towards the edges, following a linear gradient. Many seeds in the soil seed bank came from the surrounding matrices but few rainforest seeds has been dispersed to the surrounding matrices.

Edges have become ubiquitous phenomenon associating with rainforests. Through the edges, fluxes of matter, energy and species from surrounding matrices flows into the

140

rainforest, resulting in changes in species composition, structure and many ecological processes within the remaining rainforest (Laurance 1991a; Murcia 1995; Harper et al. 2005). The abrupt nature of Australian rainforest edges, mainly adjacent to pasture or eucalypt forest, has long been recognized as a key feature of Australian rainforest (Smith & Guyer 1983; Floyd 1990; Harrington & Sanderson 1994). This dramatic contrast in composition and structure between rainforest and its surrounding matrices has been proposed as the reason for the large magnitude of and long distance of edge effects, and their depth of penetration, into rainforests (Harper et al. 2005).

7.4.1 Edge-regulated soil seed bank

The species composition and seed accumulation in rainforest soil seed banks is largely determined by the magnitude of, and time since, local disturbance (Whitmore 1983; Hopkins & Graham 1984; Young et al. 1987; Abdulhadi & Lamb 1988; Garwood 1989). The seed density in soil seed banks of rainforest edges in the present study 2 ranged from 700 to 2127 / mP ,P which is higher than that found in the nearby 2 undisturbed rainforests (Abdulhadi 1989, 556 and 603 seeds/mP ,P respectively); but that of the eucalypt/ rainforest and roadside edges were comparable with the rainforest 2 reference plot in this study (918 seed/ mP ,P Chapter 5). The species richness and seed density in the soil seed bank of the pasture/ rainforest edge was close to that of a 20 year-old secondary forest, the most diverse and seed abundant soil seed bank recorded by Abdulhadi and Lamb (1988).

The rainforest soil seed banks near the edges received a considerable amount of seeds from surrounding matrices, especially from the pastures. Those generally non- rainforest species tended to be edge-specific and exhibited different distance effects according to the edge types. This concurs with the suggestion of Harper et al (2005) that the magnitude and distance of edge effects are determined by the contrast in structure and composition between the two communities that form the edge. This also highlights the necessity for studies of different rainforest edges.

7.4.2 Resilience of rainforest to disturbance

In spite of the long history of fragmentation imposed by geology and topography and,

141

more recent intensive human induced changes, the subtropical rainforest in Lamington National Park has shown considerable resilience to anthropogenic disturbance, as indicated by the composition and structure of the soil seed bank.

Apart from historical aboriginal use, early rainforest exploitation, such as cedar collecting, might have had some impact on at least some part of the national park area (Jarrott 1977). The most significant disturbance of this rainforest dates from the establishment of cattle farms and subsequent development by the O’Reilly family (Hopkins 1975). After settlement, extensive rainforest clearing was carried out and, subsequently, large tracts of secondary forests of varying ages have been created after the abandonment of some part of the clearings (Hopkins 1975). The soil seed banks under these secondary forests differ substantially from that of the nearby undisturbed rainforests (Abdulhadi & Lamb 1988).

In addition, weed seeds have become dominant components in those soil seed banks which might have long term impact on the regeneration of rainforest (Abdulhadi & Lamb 1988). According to early observations (O'Reilly 1944), some of the weed seeds might have already existed in the soil seed bank before forest clearing. For example, the ink weed, Phytolacca octandra, and cape gooseberry, Physalis peruviana, were very common on the cleared land immediately after burning. Meanwhile, rainforest clearing and subsequent abandonment of some cleared land had also promoted many rainforest pioneer species, such as Rubus rosifolius, Duboisia myoporoides and Solanum spp, which are also common elements in the soil seed bank (Hopkins 1975; Abdulhadi 1989; Olsen 1990). Moreover, the sealing of the Lamington National Park road in the late 1980’s and subsequent tourism-related development may have profoundly impacted upon the local biodiversity (Frost 2004).

All these disturbances have been suggested as having large impacts on the species composition and seed accumulation in soil seed banks under rainforest (Hopkins & Graham 1984; Garwood 1989; Hopkins et al. 1990a). Results from this study, however, suggest that the soil seed banks of the rainforests have been well maintained within rainforest interiors. This indicates a considerable resilience of local rainforest to anthropogenic disturbances and can be discussed on local and regional scales.

142

At a local scale, the rainforest edges respond actively to edge effects by changes in species composition as well as structure (Chapter 3, Chapter 4). These well developed edges can function as physical as well as biological barriers protecting rainforest interiors (Janzen 1983; Willson & Crome 1989; Murcia 1995). The distances over which soil seed banks change in species composition and seed density near edges are generally short (within 50 m). The soil seed bank at the eucalypt forest/ rainforest and roadside edges maintained a very high proportion of woody seed density (71% and 61%, respectively), which has been suggested to be a characteristic of soil seed banks under primary rainforest (Garwood 1989).

In addition, the density of exotic seeds in the soil seed bank within the eucalypt forest/ rainforest and roadside rainforest edges was comparable with that of the undisturbed rainforest. In comparison, weed seeds had penetrated further into the rainforest at pasture/ rainforest edges due to the high seed production in the pastures. Nevertheless, the weed seed density at 100 m into the pasture/ rainforest edge was comparable to that at other edges as well as to the reference rainforest. Many exotic species found at pasture edges did not occur at the other two types of edges nor in the reference rainforest even though some of sites were within a few hundred metres of the pastures. The rainforest edges in this study are all over 50 years old and well “sealed” by increases in small sized trees, especially those of edge-promoted shade-intolerant species (Chapter3). The distribution of seeds in soil seed banks in the present study has demonstrated the function of edges in maintaining the integrity of the rainforest soil seed bank.

Road construction has been identified as one of the major causes of rainforest destruction in many regions (Forman & Alexander 1998; Spellerberg 1998; Laurance & Peres 2006). Existing roads may act as important corridors facilitating the spread of weeds (Clifford 1959; Parendes & Jones 2000; Gelbard & Belnap 2003; Goosem & Turton 2006). Lamington National Park Road is probably a rare case given that the removal of large trees was minimized during road construction (Jackson & Stephenson 1986). This minimized canopy openness encouraged subsequent closure by lateral expansion of canopy trees near the edges, and has made the road alignment almost undetectable from aerial photographs. The winding road reduces the traffic speed thus lowering the impact from the wind turbulence generated by passing

143

vehicles. As a result, the changes in soil seed banks near the road are small when compared with other types of edges. The species composition and seed accumulation in soil seed bank is comparable with the adjacent undisturbed rainforest and there is no sign of weed invasion.

At a regional or even broader continental scale, the profound contrast of rainforest to other vegetation types in terms of species composition and structure may have helped maintain the integrity of soil seed banks in Australian rainforests. One of the prominent features of rainforest fragmentation in Australia is that most of the rainforest edges are surrounded by sclerophyllous forests, a vegetation type almost completely unrelated floristically to rainforest (Webb 1978; Adam 1992; Bowman 2000). Apart from frequent contraction and expansion of rainforest during a long geological history (Kershaw 1981; Hopkins et al. 1993; Hopkins et al. 1996), local fire regimes have been found to be the driving force maintaining the fragmented rainforest landscapes (Gill 1981; Ash 1988; Unwin 1989; Adam 1992; Bowman 2000). Consequently, many plant species from sclerophyllous forest need fire related cues to initiate germination of soil stored seed (Dixon et al. 1995; Enright et al. 1997; Read et al. 2000; Enright & Kintrup 2001); these species are not likely to interfere with the regeneration of rainforest as fires are generally not common within rainforest (Floyd 1990; Adam 1992). Meanwhile, many sclerophyllous trees can retain seeds over a long period by forming a canopy seed bank until released by fire (Lamont et al. 1991; Enright et al. 1998). In addition, many species from sclerophyllous forest have adapted to dispersal by ants which can only move seeds over short distances (Berg 1975; Hughes & Westoby 1992b, a). As a result, few sclerophyllous species can be found in rainforest soil seed banks adjacent to eucalypt forests (Wang 1996; Tang et al. 2003).

Outside Australian, long term shifting agriculture have had a profound impact on the species composition, structure and dynamics of rainforest (Ewel et al. 1981; Uhl et al. 1981; Fox et al. 2000; Willis et al. 2004). The cultivation and subsequent regeneration of fallows can significantly promote short-lived rainforest pioneers as well as agricultural weeds and, as a result, lead to dramatic changes in species composition and seed accumulations in soil seed banks under regional rainforest (Young et al. 1987; Saulei & Swaine 1988; Garwood 1989; Cao et al. 2000). Many of the disturbance-

144

induced species from genera such as Trema, Macaranga, Melastoma and Anthocephalus in Southeast Asia (Cheke et al. 1979; Whitmore 1983; Saulei & Swaine 1988; Cao et al. 2000; Tang et al. 2006), Cecropia, Micronia and Vismia in Amazon (Uhl et al. 1981; Uhl 1987), are also common early occupants during gap dynamic of mature rainforest (Putz 1983; Whitmore 1983; Denslow 1987; Putz & Appanah 1987; Uhl et al. 1988). The dynamics of rainforest edges have been largely affected by these successional species (Didham & Lawton 1999; Laurance et al. 2006b).

In Australia, however, aboriginal people have never practiced shifting agriculture (Cosgrove 1996; Bowman 2000; Hill et al. 2000). Although a long history of using fire by aboriginal people may have substantially changed the rainforest landscape, most of those fire events, however, happened in vegetation types other than rainforest (Lesley 1989; Fensham 1997; Bowman 2000). These fire-induced disturbances may cause frequent contraction and expansion of rainforest edges, but no large area of secondary rainforest forest will be generated as the burnt areas will be normally claimed by sclerophyllous species (Floyd 1976; Smith & Guyer 1983; Ash 1988; Unwin 1989). This different land use history and contrast in species composition has determined that there is limited seed exchange between rainforest remnants and surrounding vegetation. As a result, historical human activities might have had much less impact on Australian rainforests than they had on other rainforest area.

In addition, the absence of large-sized soil-disturbing animals such feral pigs, may have also contributed to the maintenance of a rainforest soil seed bank in a large rainforest remnant (20,600 hectares) like Lamington National Park. Large sized destructive animals may have devastating impact on rainforest fragments by causing direct damage and, more seriously, facilitating the deep penetration of hostile edge effects (Laurance 2000; Peters 2001). In North Queensland, feral pig proliferation has caused considerable environmental damage in rainforest (Pavlov et al. 1992; Laurance & Harrington 1997). Within the study area, the red-necked Pademelon, Thylogale thetis, may cause substantial impacts on rainforest regeneration by selective browsing rainforest seedlings (Wahungu et al. 1999). This impact, however, has mostly been confined within few metres of rainforest edges and mainly on pioneer species (Wahungu et al. 1999).

145

7.4.3 Dispersal limitation of rainforest seeds in fragmented landscapes

Rainforest remnants have been considered as important seed sources that may facilitate the natural regeneration of forest in surrounding deforested areas (Turner & Corlett 1996; Aide et al. 2000; Martinez-Garza & Howe 2003; Lamb et al. 2005). Strong seed dispersal limitation, however, has been found to be a major constraint to the fulfillment of this function and hampers the restoration of deforested areas (Ferguson & Drake 1999; Holl 1999; Wijdeven & Kuzee 2000; Cubina & Aide 2001; White et al. 2004; Hooper et al. 2005). In this study, the seed density of tree species showed a linear decrease from rainforest interior towards edges. Furthermore, there was not a single native rainforest tree seed germinated from the soil sample of the pastures despite the fact that some of the sampling spots were just a few metres away from the rainforest edge. This implies that there is likely to be a strong seed dispersal limitation between the rainforest and pastures.

Lack of suitable perch site and food sources have been identified as one of the major obstacles limiting the dispersal of rainforest seed into deforested areas (Willson & Crome 1989; Da Silva et al. 1996; Ferguson & Drake 1999; Toh et al. 1999). Even isolated trees in deforested area, especially fleshy-fruited species, may significantly increase the seed dispersal events and hence promote nature regeneration (Guevara et al. 1986; Otero-Arnaiz et al. 1999; Toh et al. 1999; Galindo-Gonzalez et al. 2000). The occurrence of some pioneer tree seeds in soil seed banks under eucalypt forests provided evidence of the importance of perch site in seed dispersal. This hypothesis, however, cannot fully explain the seed distribution patterns found in current study. First, the linear decrease of the density of rainforest tree seeds in soil seed banks started from about 100 metres into the rainforest, at which point on the composition and structure of the forest was similar to that of undisturbed forest. Many trees and lianas near the edges bear fleshy fruits that may provide food sources for fruigivores (Chapter 3, 4). Similarly, Holl et al (1998) found the existence of seed dispersal limitation even in habitats with perch sites and food sources provided. Second, in the current study, many species that had seed density negatively correlated with the distances from rainforest interior to edge, such as C. paniculosa and P. lanchnocapa, are actually wind dispersed and hence their distribution is not affected by frugivore

146

activities.

Well developed rainforest edges can become effective physical barriers protecting the forest interior from exposure to the generally hostile matrices (Janzen 1983; Williams- Linera 1990b; Laurance 1991b; Murcia 1995; Nascimento et al. 2006). At the same time, the edges may also act as functional screens reducing the flux of matter, such as nutrients, pollutants, seeds and even animals, from the surrounding matrices into the rainforest (Cadenasso & Pickett 2000; Cadenasso & Pickett 2001; Weathers et al. 2001; Devlaeminck et al. 2005). In an experiment comparing the seed flow across thinned and intact edges, Cadenasso and Pickett (2001) have demonstrated the function of edges as barriers preventing wind-dispersed seed penetrating into forest interiors. The density of seeds originating from pasture in the present study, many of them wind dispersed, showed a sharp drop from pasture to rainforest and a continuing decrease towards the rainforest interior. This indicates the existence of such a seed dispersal barrier. What Cadenasso and Pickett did not state is that this barrier can also work in the opposite direction limiting the dispersal of rainforest seed into the surrounding matrices. The present study found apparently two opposite directions of seed movement near the edge reflecting the origin of seeds in the soil seed bank. Seeds originating from pasture or eucalypt forest showed a decrease from the edge to rainforest interior whereas rainforest tree seeds showed increases in the same direction. Even the roadside edges showed an obvious decrease of tree seeds from the rainforest interior toward the edges. This gradient may be due partly to the wind turbulence generated by passing vehicles which created the same result as wind at open edges. This showed that edge structure, with the interaction with wind, plays an important role in seed movement across rainforest edges. It is crucial, therefore, to keep the integrity of rainforest edges in order to protect rainforest soil seed banks and reduce weed invasion.

Many other factors may also account for the absence of rainforest seeds in nearby deforested areas even where seed dispersal has occurred , such as seed predation (Nepstad et al. 1996; Wijdeven & Kuzee 2000) and infections by fungal pathogen (Blaney & Kotanen 2001). Moreover, the harsh environmental conditions such as high radiation levels, strong temperature fluctuations and frequent soil disturbance may not be suitable for rainforest seed to persist in soil (Uhl et al. 1981; Garwood 1989;

147

Vazquez-Yanes & Orozco-Segovia 1993; Williams-Linera et al. 1998). Inevitably, these negative effects may also penetrate the edge and cause changes in soil seed banks within rainforest (Williams-Linera 1990b, a). Even if, occasionally, some seed do arrive in the deforested areas they may germinate and then die due to high herbivory pressure or lack of establishment requirements (Nepstad et al. 1996). Together with the seed dispersal limitation discussed above, these factors may limit the accumulation of viable rainforest seeds in the soil seed bank and, therefore, suppress the natural regeneration of the deforested area. Direct seeding or planting has been suggested to overcome this apparent ecological threshold for natural rainforest recovery (Aide et al. 2000; Lamb et al. 2005).

7.4.4 Conservation applications

This study identified the pastures as the major sources of exotic species in the soil seed bank; hence, the management of weeds in pasture is important to maintaining the integrity of rainforest soil seed banks. Conversion of pastures to forest, especially those in the core area of the National Park, should be considered. Meanwhile, the present study also highlights the importance of maintaining the integrity of existing rainforest edges. In addition, maintaining existing secondary forests can be crucial to overcoming seed dispersal limitation and, in turn, encouraging natural forest regeneration.

148

8 GENERAL DISCUSSION

8.1 Summary of Findings

This study has used extensive vegetation surveys and germination experiments to investigate the composition and structure of assemblages of canopy trees and lianas, seedling banks and soil seed banks across three types of rainforest edges. These findings were compared with results from a rainforest reference plot in a relatively undisturbed area. The results have shown that the plant species composition and structure of the rainforest, including trees and lianas, have been largely modified at edges. These modifications, interacting with the disturbances from the surrounding matrices, have led to significant changes in seedling and soil seed banks which, in turn, may have long term impacts on the dynamics of local and regional rainforests.

8.1.1 Changes in the rainforest canopy

In terms of forest structure, the most noticeable change at rainforest edges is the increase in the density of small trees and small lianas from the rainforest interior to the edges (Chapter 3 and Chapter 4). The densities of small trees and small lianas were significantly correlated with each other and presumably reflect their similar responses to the increased light regimes near the rainforest edges (Chapter 4). The distances over which the changes in forest structure can be detected varied among different types of edges, ranging from within 10 m at roadside edges to around 80 m at pasture/ rainforest edges. To a large extent, these changes in forest structure were related to the openness of rainforest canopies at different types of edges, with, for example, the relatively open canopy at pasture edges permitting more light to illuminate the forest floor, which in turn has promoted the growth of small trees and lianas. The increase in the density of small trees and lianas may reduce the permeability of rainforest edges and help protect the rainforest interior from exposure to negative edge effects.

The species compositions of the forests have also been changed significantly at rainforest edges when compared with the reference rainforest. These changes include an increase in the density and/ or proportion of shade-intolerant trees (Chapter 3) and

149

lianas using tendril for climbing and lianas dispersed by frugivores (Chapter 4). Different suites of species (Chapter 3 and Chapter 4) characterized different types of edges.

8.1.2 Changes in the soil seed bank and seedling bank

The changes in forest canopies at rainforest edges have also led to significant changes in the seedling and soil seed banks across all three types of edges (Chapter 5, Chapter 7). The seedling banks, which rely heavily on the microclimatic conditions maintained by the forest canopies, showed different distribution patterns from edges to rainforest interiors depending on the type of edge (Chapter 5). The seedling banks showed an increase in the density of shade-tolerant species from the edge to rainforest interiors and a corresponding decrease in shade-intolerant species. The distributions of seedling banks in relation to the species composition and structure of the forest canopy were found to be different between the reference rainforest and edges, and among edges themselves (Chapter 5). Compared with the reference rainforest, a significant decrease in the density of seedlings of mature phase rainforest species was found at eucalypt forest/ rainforest and pasture/ rainforest edges, with a corresponding increase in the density of lianas. In addition, the roadside edges also showed a significant higher density of liana seedlings than the reference rainforest despite the fact that they had a similar density of tree seedlings. This implies that edge effects may have favoured the establishment of liana seedlings at the expense of tree seedlings at these edge types, providing strong evidences for the notion that edges promote liana proliferation (Laurance et al. 2001b; Schnitzer & Bongers 2002).

The species composition and seed density in the soil seed banks have also been modified substantially at the rainforest edges when compared with that of the reference rainforest (Chapter 5 and Chapter 7). The most significant changes are a decrease in the density of rainforest seeds and an increase in both the number of species and seed density of non-rainforest seeds originating from the surrounding matrices, principally agricultural weeds, at rainforest edges compared with reference rainforest. These changes show strong interactions between the rainforest and surrounding matrices and demonstrate the function of rainforest edges as an effective barrier protecting the rainforest interiors against the dispersal of agricultural weeds

150

(Chapter 7). At the same time, however, the edges may also cause dispersal limitation preventing rainforest seeds being dispersed into surrounding areas - which may hinder natural regeneration in deforested areas. This exchange of species between rainforest and the surrounding matrices can have important implications in rainforest conservation as well as restoration.

8.1.3 Edges under different disturbance regimes

Pasture/ rainforest edge The changes in the species composition and structure of rainforest at pasture/ rainforest edges are largely related to the opening of the rainforest canopy and subsequent changes in microclimatic factors such as light, temperature and wind speed during the early stage of edge creation (which typically involves the clear- felling and burning of the forest trees). The species that respond to these changes are mostly those that can also be found in rainforest gaps. Because of the contrast in structure between the rainforest and pasture, these changes can penetrate deeper into the rainforest than at the other two types of edges. The results from this study showed that pasture/ rainforest edges are undergoing vigorous secondary succession. These changes are similar to those found at pasture/ rainforest edges in other rainforest areas such as the Amazon (Laurance et al. 1998a; Williams-Linera et al. 1998; Laurance et al. 2001b; Laurance et al. 2006b), Mexico (Williams-Linera et al. 1998) and Costa Rica (Schedlbauer et al. 2007).

Large numbers of agricultural weeds as well as pioneer rainforest species have been promoted by cattle farming. This has led to a dramatic increase in their seed density in the soil seed bank near the edges. Although most of the weed species were only found in the soil seed bank of pasture/ rainforest edges in this study and there was no sign of their spreading into other edges or the rainforest reference plot, the presence of these weeds still should be a management concern as these weeds may potentially spread into disturbed such as rainforest gaps, given that the pastures are located close to the core area of the National Park.

151

Eucalypt forest/ rainforest edge The eucalypt forest/ rainforest edges, in contrast, have been subjected to long-term edge effects, especially repeated burning (Chapter 1). The edges have been well sealed by a layer of almost impenetrable small trees and lianas, many of which may be considered to be fire adapted (Chapter 3 and Chapter4). The structure of this kind of edge, and the associated level of disturbance is comparable, to some extent, to the gallery forest edges in central America (Kellman et al. 1996). The contrast in species composition between the two forests is also reflected into the soil seed banks in which few species of rainforest species can be found in the soil seed bank of eucalypt forest and vice versa. This has shown that a strong dispersal limitation may exist between the rainforest and the eucalypt forest, or that conditions promoting the persistence of seeds in the soil seed bank vary markedly between the two forest types. In places, where the expansion of rainforest is desirable, human intervention such as seed addition and soil disturbance may be needed to encourage rainforest regeneration.

Roadside rainforest edge Despite the significant increase in the density of small trees and lianas, the species composition of roadside rainforest edges is generally similar to that of the rainforest interior (Chapter3 and Chapter 4). No obvious signs of invasion of shade-intolerant species, such as occurred at the other two types of edges, were found, mainly due to the limited canopy openness associated with the narrow road clearing. The increase in the density of small trees may have been a result of release of the existing seedling bank following edge creation. The lianas’ proliferation at the roadside edges is caused mainly by ramet proliferation of existing parent plants and invasions by lianas from the interior of the forest through horizontal movement, possibly induced by the increase in lateral light penetration. To a large extent, these changes are similar to that found within small forest gaps and can only be detected a few metres into the rainforest.

Even though there is no strong sign of weed infestation along the roadside edges, the higher light environment has provided habitats for many shade-intolerant species (Plate 2). Many of these species have already developed a considerable accumulation of seeds in the soil seed bank, for example Hydrocotyle pedicellosa and Nyssanthes diffusa. There are also many large size canopy trees of Dendrocnide excelsa growing

152

along the roadside edges. It is likely that the regeneration and growth of this pioneer species has been promoted by the road construction and has provided a large amount of seed in the soil seed bank (Chapter 7). How these changes in both above-ground vegetation and soil seed bank affect rainforest regeneration near the edges needs further study and should draw the attention of the management authorities, particularly when alterations to the existing road network (e.g. widening existing roads) are being considered.

8.2 The Impact of the Prolonged Drought

The study area has been subjected to a prolonged drought since 2000 (for detailed description see Chapter 1), which may have had a strong impact on the dynamics of the rainforest and thus the results of the current study.

Periodical drought can change the phenological behaviour of many rainforest species including causing synchronized mast fruiting (Ashton et al. 1988; Kelly & Sork 2002; Williamson & Ickes 2002). Large cohorts of rainforest seedlings may be generated after the mast fruiting. This has been proposed as a strategy to satiate natural enemies, both seed predators and seedling herbivores (Janzen 1974; Whitmore 1998a; Curran & Webb 2000). On the other hand, prolonged drought may also increase seedling mortality rates due mainly to water stress during the drought (Gilbert et al. 2001; Bebber et al. 2004). Even though studies on tree mortality from Amazonian rainforest have showed that drought may not interact directly with the edge effects (Laurance et al. 2001a), results from the present study suggest that the prolonged drought may have had strong impacts on the soil seed bank and seedling banks by, possibly, inducing mast seeding and alteration of microclimate conditions that affect seedling establishment (Chapter 5, Chapter 6, Chapter 7). In addition, it is possible that there is an interaction of edge effects and the drought, affecting the distribution of seeds and seedlings along the edges.

This study was not designed specifically to investigate the impact of the drought and there was no comparable investigation before the drought. The seeds of the canopy tree Caldclavia paniculosa in the soil seed bank and seedlings of the vine Derris involouta in the seedling bank, however, showed strong evidence that a drought

153

induced mast fruiting and associated changes in microclimatic conditions have had substantial impact on the distribution of soil seed banks and seedling banks (Chapter 5, Chapter 6). The negative associations between liana and tree seedlings found at rainforest edges in the present study may be a result of the interaction of edge effects and drought as lianas have been suggested to be more competitive than trees during drought (Schnitzer 2005). The long term impact of these changes on the dynamics of local rainforest needs further investigation; in particular, the large cohort of seedlings of D. involuta, which may result in liana proliferation near rainforest edges, requires long term monitoring.

8.3 Using Soil Seed Bank as Indicators of Edge Effects

Apart from direct measurement of physical environmental factors, which normally become insignificant or even disappear as the edges age (Saunders et al. 1991; Kapos et al. 1997), many biological variables have been used as indicators for the magnitude and penetration of edge effects (Murcia 1995; Laurance et al. 2002; Harper et al. 2005). These abiotic and biotic factors always interact with each other and can lead to changes in rainforest composition and structure from a few metres to hundreds of metres from edges into rainforest interiors, depending on the variables of concern and the magnitude of, and the time since, disturbance (Williams-Linera et al. 1998; Laurance et al. 2002; Harper et al. 2005). In the present study, trees, lianas, seedling banks and soil seed banks, have all shown strong but different responses to the edge effects. The changes in trees and lianas, however, may be largely related to the initial disturbance that generated the edges and may not reflect current edge effects. Seedling banks may be more sensitive to edge effects than trees and lianas but rainforest seedlings can also survive for tens of years which may require long-term monitoring to estimate the changes. Those well established plants may show considerable resilience to edge effects thus their response to fragmentation effects may show a strong time lag (Tilman et al. 1994; Vellend et al. 2006). A snap shot of all of the above three elements may reflect, more or less, the structural complexity, but not the species composition of the surrounding matrices, as most of the species in the surrounding matrices are generally absent in the rainforest.

Soil seed banks are highly dynamic and have been found to contain information of

154

past disturbance in rainforests as well as having the ability to predict future regeneration (Saulei and Swaine 1988, Garwood, 1989, Dalling et al, 1998). The composition and structure of soil seed banks may also reflect local and even regional disturbances (Hopkins & Graham 1984; Hopkins et al. 1990a; Cao et al. 2000; Lin et al. 2006). As demonstrated in a edge manipulation experiment (Cadenasso & Pickett 2001), the movement of seeds across rainforest edges are largely affected by the structure of the edges itself. The changes in microclimatic conditions at the edges may also cause significant changes in the soil seed bank, such as inducing germination or causing death (Williams-Linera 1990a). The species composition and seed distribution in soil seed banks is also affected largely by ecological processes such as pollination, seed dispersal and predation (Willson & Crome 1989; Nepstad et al. 1996; Wijdeven & Kuzee 2000). To some extent, therefore, soil seed bank may be considered as the end product of many abiotic and biotic edge effects. The study of soil seed banks at rainforest edges may well capture the short-term as well as long-term changes in the physical and biological factors across the two communities that form the edge.

The soil samples collected from rainforest edges in the present study contain species from both rainforest as well as surrounding matrices. Many non-rainforest seeds are generally edge specific and their distribution in the soil seed bank can be strongly related to the composition and structure of the edges and their interactions with the surrounding matrices (Chapter 7).

The study of the soil seed bank at edges may also provide important information for rainforest restoration, such as identifying restoration thresholds (Wijdeven & Kuzee 2000; Cubina & Aide 2001). Seed dispersal seems to be limited at rainforest edges which may relate to the changes in the activities of seed dispersers in response to different edge types (Chapter 7). This will have important implication in restoration, for example, seed direct seed addition or creating frugivore habitat, to promote natural regeneration.

Using soil seed banks as indicators may also provide early-warning of weed infestation, a common consequence of rainforest fragmentation. In this study, many exotic species that did not occur in the above-ground vegetation have been found in the soil seed bank. Some of them, such Ageratina riparia and Senecio

155

madagascariensis are weeds that may not grow in intact rainforest but potentially can interact with rainforest regeneration in disturbed areas such as the roadside. The study of the soil seed bank can provide important information for monitoring the movement of these weed species in a landscape.

8.4 Regeneration Strategy of Rainforest Plants

Classifying rainforest plants into ecological groups has been widely accepted in rainforest studies and has become one of the fundamental approaches to understand the dynamics of rainforests (Hopkins et al. 1977; Swaine & Whitmore 1988; Whitmore 1989; Dalling et al. 2004; Laurance et al. 2006b). It has also been adopted in rainforest restoration, for example, using pioneer species to initiate regeneration and using primary species to accelerate succession (Floyd 1990; Martinez-Garza & Howe 2003). Many such classification systems are based mainly upon field observations of plant life history and regeneration requirements rather than ecological and physiological studies (Hopkins 1975; Swaine & Whitmore 1988). In Australia, ecological grouping have frequently derived from studies of rainforest regeneration (Webb et al. 1972; Hopkins 1975; Shugart et al. 1980; Hopkins & Graham 1987; Stewart 1995; Osunkoya 1996). These classifications have been based mainly on individual studies or observations from particular sites and their conclusions may differ considerably from each other. Many species classifications are uncertain due to a lack of relevant ecological and physiological information.

In the present study, the soil seed bank and the seedling bank are composed of different sets of species with only a slight overlap. This generally supports the dichotomous classification of rainforest plants into two broad ecological groups based on their regeneration strategies; ‘pioneer’ species, which rely mainly on the soil seed bank, and ‘climax’ species which depend primarily on seedling banks for regeneration (Hopkins & Graham 1987; Swaine & Whitmore 1988; Whitmore 1989). Disturbances within the rainforest may allow the germination and short term appearance of some pioneer species within the seedling bank. Primary rainforest species may also be present transiently in the soil seed bank before germination, for example Atractocarpus benthamianus in the present study. In addition, a few small-seeded shade-tolerant species are found in many rainforest, apparently adapted to

156

germinating in litter-free sites (e.g. steep slopes, Metcalfe & Turner 1998). This may result in the co-occurrence of some species in both the soil seed bank and the seedling bank.

The subtropical rainforest in the study area contains elements of both tropical and temperate origins (Burbidge 1960). One of the significant features of this kind of subtropical rainforest is the occurrence of many long-lived pioneer species, such as Caldcluvia paniculosa and Geissois benthamii, which may become dominant in both the tree canopy and the soil seed bank. These long-lived shade-intolerant species can have important applications in rainforest conservation and restoration (Chapter 6). The traditional classification system by Swaine and Whitmore (1988) may not capture this significance. In this study, I adopted the classification system of Shugart et al (1980) using shade-tolerant, intermediate and shade-intolerant as three ecological guilds but further classified shade-intolerant species into short-lived and long-lived to emphasize the long-lived pioneer species. This classification is based on the species occurrence in the soil seed banks, seedling banks and tree canopy (Table 8.1). To some extent, this approach is similar to that developed by Osunkoya (1996), for tropical Australian rainforest species, based on the size of forest gaps that a species requires for successful regeneration. Due to limited data on most of the species, this classification is no more than just a conceptual model at this stage with some examples from the present study. A similar attempt has been made by Stewart (1995) , for Australian subtropical rainforests, based on seed and seedling studies in a secondary rainforest.

Short-lived shade-intolerant species may have many persistent seeds in the soil seed bank. Their seeds normally cannot germinate under an dense canopy due to limited radiation. As a result, they are generally absent from the seedling bank under intact canopy although they may occasionally be found in recently opened light gaps. These normally fast-growing and short-lived small trees are likely to be shaded out rapidly after the closure of the forest canopy. Only after large areas of rainforest destruction, such as cyclone damage, or large areas of clearing, may those species be present as a large cohort during the early stage of secondary succession (Olsen and Lamb 1989, Hopkins, 1977). Typical representatives of this group include species such as Solanum mauritianum, Duboisia myoporoides and Trema tomentosa. This group is similar to

157

group ‘A2’, early secondary species, in Hopkins’s (1975) classification system and is classified as shade-intolerant in Shugart et al. (1980). These species have been frequently used to initiate rainforest restoration (Floyd, 1990).

Long-lived shade-intolerant species dominate the soil seed bank in this sub-tropical rainforest (Chapter 3). Many mature individuals are present as large canopy trees and act as consistent seed sources. The seeds may persist in the soil seed bank for a relatively long period of time but not germinate under the low light environment of the intact canopy. The conditions in forest gaps may be sufficient to initiate seed germination and seedling establishment of these species. Typical species include Caldcluvia paniculosa and Geissois benthamii. This group is similar to Group ‘B’ in Hopkins’s system and shade-intolerant species in the classification scheme of Shugart et al. (1980).

The two groups, short- and long-lived shade-intolerant species, described above may be considered as pioneers in the dichotomous grouping of Swaine and Whitmore (1988).

Intermediate species can have small but persistent seed accumulations in the soil seed bank. Despite preferring light gaps, this group may also germinate and persist in the seedling bank under an intact canopy and can frequently be found in the canopy. Species in this group include Pseudoweinmannia lachnocarpa, and many species from the family Rutaceae. In Hopkins’ system, these species were mainly considered as late secondary species. As pointed out by Swaine and Whitmore(1988), the variation from pioneer to primary, or from shade-intolerant to shade-tolerant, is continuous. Any further division would be arbitrary.

Shade-tolerant species are generally absent from the soil seed bank. Their seeds tend to be relatively large and have evolved fast germination strategies (Hopkins & Graham 1987; Vazquez-Yanes & Orozco-Segovia 1993). Seedlings of those species have the ability to survive beneath the tree canopy and are the main component of the rainforest seedling bank. Trees of shade-tolerate species are the main components of mature phase rainforest.

158

Table 8.1 Classification of tree species according to their occurrence in the soil seed bank, seedling bank and canopy trees. (This classification scheme is modified from Shugart et al (1980), based on the light requirement of a tree species for a successful regeneration. Similar classification using plant presence in different successional stages was proposed by Hopkins (1977), in the same type of forest as the present study).(- Few/ absent, * Common, ** Many; for more species see Appendix 1) Description Soil seed Seedling bank Canopy Example bank Short-lived shade- ** - - Duboisia myoporoides, Polyscias elegans, Rubus intolerant species rosifolius, Solanum mauritianum, Solanum aviculare, Trema tomentosa, Dendrocnide excelsa, Long-lived shade-in- ** - * Caldcluvia paniculosa, Geissois benthamii tolerant species Intermediate species * * * Acronychia suberosa, Acronychia octandra, Pseudoweimannia lachnocarpa Shade tolerant species - ** ** Actephila lindleyi, Argyrodendron spp, Atractocarpus benthamianus, Baloghia inophylla, Diospyros pentamera

159

There are also many typical shade-intolerant species, such as Homalanthus poulifolius and Alphitonia excelsa that did not appear to be well-represented in the soil seed bank, at the sites sampled in this study. This may reflect characteristics of the study site (e.g., paucity of mature trees of these species at study sites), or attributes of the seeds themselves. For example, some pioneer species such as Alphitonia and Acacia may need heat shock or other dormancy-breaking treatments to initiate germination. Consequently, germination using untreated soil samples (the method used in the present study) may underestimate the presence of such species in the soil seed bank. A reliable classification of rainforest trees and shrubs by regeneration guild is still to be developed and will need long term intensive ecological and physiological studies.

8.5 Rainforest Edge: Future for Australian Rainforest

Promoting natural regeneration has been considered as one of the most cost effective ways of restoring rainforest to cleared or degraded land (Aide 2000; Erskine 2002; Lamb et al. 2005). Fleshy-fruited pioneer species that are capable of long distance dispersal by frugivores play crucial roles in this process, especially during the early stages of natural regeneration (Aide 2000). As discussed elsewhere in this thesis, a lack of large areas of secondary forest may prove to be an obstacle for large-scale natural rainforest regeneration in Australia, due to the unique historical rainforest fragmentation background and the recent rapid and intensive rainforest destructions. Rainforest edges may become one of the few important seed sources for natural regeneration by supporting large amounts of flesh-fruited pioneer trees and lianas. The existence of dispersal limitation at edges, however, may compromise this function. Working out how to overcome this dispersal limitation to fulfill the function of edges as seed sources is a priority, in both rainforest conservation and future restoration.

8.6 Future Research Needs

This study has shown that the magnitude and penetration distance of edge effects, in the area of subtropical rainforest examined, are largely affected by the characteristics of surrounding matrices and their associated disturbance regimes. Consequently, some of the conclusions of this study are largely limited to this particular study area. Further studies of different types of edges at a landscape scale are needed for a more

160

comprehensive understanding of the impact of edge effects and the applications in rainforest conservation.

Although the studies described in this thesis show that the soil seed bank could potentially be used as a tool to monitor the conditions and long term dynamics of rainforest edges, more studies are needed before any practical applications of this suggested approach can be implemented. For example, due to the difficulties of seedling identification, simplified procedures using functional guilds could be considered to develop standardized rapid assessment techniques, instead of relying on the identification of the entire soil seed bank flora. A similar approach might also be adopted in other areas requiring rapid assessment of rainforest dynamics, such as the planning and assessment of rainforest restoration (Kanowski & Catterall 2007).

The present study has provided a baseline for the study of the impact of the prolonged drought on the local rainforest dynamics as well as for study of the interactions between drought and edge effects. The monitoring of the large seed accumulation of C. paniculosa in the soil seed bank and the cohort of seedlings of D. involuta in the seedling bank may generate significant results regarding the dynamics of the local rainforest after the drought.

Long-lived pioneer species, which can accumulate large number of seeds in soil seed bank and which also can become dominant in rainforest canopies, may play important roles in the dynamic of subtropical rainforest. This provides a unique opportunity to test the hypotheses of recruitment limitation (Hubbell et al. 1999) and density dependence (Janzen 1970; Connell 1971), regarding the origin and maintenance of rainforest diversity. Studies of the dynamics of these long-live pioneer species may also have significant applications in rainforest conservation as well as restoration. For example, long-lived pioneers may be a particularly useful component of restoration plantings. Recent experience suggests that short-lived pioneers, which are more commonly used in restoration plantings, can allow the re-invasion of replanted sites by light-demanding grasses and weeds when the pioneers senesce (Freebody 2007)

Dispersal-limitation of rainforest seeds has been identified along the edges in this study. How this dispersal limitation affects the dynamics of rainforest edges and the

161

regeneration of deforested areas, and how to overcome this limitation to encourage natural regeneration should be one of the priorities for future researches. Manipulative experiments such as seed or seedling addition to deforested areas may be considered to identify ecological constraint to rainforest restoration.

162

REFERENCES

Abdulhadi, R. 1989. Sub-tropical Rain Forest Seed Dynamics. Thesis (Ph.D., page 300), University of Queensland.

Abdulhadi, R., and D. Lamb. 1988. Soil seed stores in a rainforest succession. Proceedings of the Ecological Society of Australia 15:81-87.

Adam, P. 1992. Australian Rainforest. Oxford University Press, Oxford.

Aide, T., and J. Zimmerman. 1990. Patterns of insect herbivory, growth, and survivorship in juveniles of a Neotropical liana. Ecology 71:1412-1421.

Aide, T. M. 2000. Clues for tropical forest restoration. Restoration Ecology 8:327-327.

Aide, T. M., J. K. Zimmerman, J. B. Pascarella, L. Rivera, and H. Marcano-Vega. 2000. Forest regeneration in a chronosequence of tropical abandoned pastures: implications for restoration ecology. Restoration Ecology 8:328-338.

Aizen, M. A., and P. Feinsinger. 1994. Forest fragmentation, pollination, and plant reproduction in a chaco dry forest, Argentina. Ecology 75:330-351.

Allen, B., E. Pauley, and R. Sharitz. 1997. Hurricane impacts on liana populations in an old-growth southeastern bottomland forest. Journal of the Torrey Botanical Society 124:34-42.

Alvira, D., F. E. Putz, and T. S. Fredericksen. 2004. Liana loads and post-logging liana densities after liana cutting in a lowland forest in Bolivia. Forest Ecology and Management 190:73-86.

Ash, J. 1988. The location and stability of rainforest boundaries in North-Eastern Queensland, Australia. Journal of Biogeography 15:619-630.

Ashton, P. S., T. J. Givnish, and S. Appanah. 1988. Staggered flowering in the dipterocarpaceae: new insights into floral induction and the evolution of mast fruiting in the aseasonal tropics. The American Naturalist 132:44-66.

Augspurger, C. K. 1984a. Light requirements of neotropical tree seedlings: a comparative study of growth and survival. The Journal of Ecology 72:777-795.

Augspurger, C. K. 1984b. Seedling survival of tropical tree species: interactions of dispersal distance, light gaps, and pathogens. Journal of Ecology 72:777-795.

Babweteera, F., A. Plumptre, and J. Obua. 2000. Effect of gap size and age on climber abundance and diversity in Budongo Forest Reserve, Uganda. African Journal of Ecology 38:230-237.

163

Bailey, R. C., G. Head, M. Jenike, B. Owen, R. Rechtman, and E. Zechenter. 1989. Hunting and gathering in tropical rain forest:is it possible? American Anthropologist 91:59-82.

Baker, S. C., L. A. Barmuta, P. B. Mcquillan, and A. M. M. Richardson. 2007. Estimating edge effects on ground-dwelling beetles at clearfelled non-riparian stand edges in Tasmanian wet eucalypt forest. Forest Ecology and Management 239:92-101.

Balee, W., and D. Campbell. 1990. Evidence for the successional status of liana forest (Xingu River Basin, Amazonian Brazil). Biotropica 22:36-47.

Balfour, D. A., and W. J. Bond. 1993. Factors limiting climber distribution and abundance in a southern African Forest. Journal of Ecology 81:93-99.

Bazzaz. 1983. Dynamics of wet tropical forests and their species strategies. Pages 233-243 in E. Meedina, H. Mooney, and C. Vazquez-Yanes, editors. Physiological Ecology of Plants of the Wet Tropics. Dr W. Junk Publishers, The Hague, Netherlands.

Bebber, D., N. Brown, and M. Speight. 2004. Dipterocarp seedling population dynamics in Bornean primary lowland forest during the 1997-8 El Nino Southern Oscillation. Journal of Tropical Ecology 20:11-19.

Becker, P., and A. P. Smith. 1990. Spatial autocorrelation of solar radiation in a tropical moist forest understory. Agricultural and Forest Meteorology 52:373- 379.

Benítez-Malvido, J. 1998. Impact of forest fragmentation on seedling abundance in a tropical rain forest. Conservation Biology 12:380-389.

Benítez-Malvido. J., and A. Lemus-Albor. 2005. The seedling community of tropical rain forest edges and its interaction with herbivores and pathogens. Biotropica 37:301-313.

Benítez-Malvido, J., and A. Lemus-Albor. 2006. Habitat disturbance and the proliferation of plant diseases. Pages 165-174 in W. F. Laurance, and C. A. Peres, editors. Emerging Threats to Tropical Forests. The University of Chicago Press, Chicago and London.

Benítez-Malvido, J., and M. Martínez-Ramos. 2003a. Impact of forest fragmentation on understory plant species richness in Amazonia. Conservation Biology 17:389-400.

Benítez-Malvido, J., and M. Martínez-Ramos. 2003b. Influence of edge xposure on tree seedling species recruitment in tropical rain forest fragments. Biotropica 35:530-541.

Berg, R. Y. 1975. Myrmecochorous plants in Australia and their dispersal by ants. Australian Journal of Botany 23:475-508.

164

Blaney, C. S., and P. M. Kotanen. 2001. Effects of fungal pathogens on seeds of native and exotic plants: a test using congeneric pairs. Journal of Applied Ecology 38:1104-1113.

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

Bowman, D. 2000. Australian Rainforests. Cambridge University Press New York, NY, USA.

Bowman, D., and W. Panton. 1993. Factors that control monsoon-rainforest seedling establishment and growth in north Australian Eucalyptus savanna. The Journal of Ecology 81:297-304.

Brokaw, N. 1987. Gap-phase regeneration of three pioneer tree species in a tropical forest. The Journal of Ecology 75:9-19.

Brokaw, N., and R. T. Busing. 2000. Niche versus chance and tree diversity in forest gaps. Trends in Ecology & Evolution 15:183-188.

Brokaw, N. L., and S. M. Scheiner. 1989. Species composition in gaps and structure of a tropical forest. Ecology 70:538-541.

Brook, B. W., C. J. A. Bradshaw, L. P. Koh, and N. S. Sodhi. 2006. Momentum drives the crash: mass extinction in the tropics. Biotropica 38:302-305.

Bruna, E. M. 2002. Effects of forest fragmentation on Heliconia acuminata seedling recruitment in central Amazonia. Oecologia 132:235-243.

Bruner, A. G., R. E. Gullison, R. E. Rice, and G. A. B. Da Fonseca. 2001. Effectiveness of parks in protecting tropical biodiversity. Science 291:125-128.

Burbidge, N. T. 1960. The phytogeography of the Australian region. Australian Journal of Botany 8:75-211.

Burgess, I. P., A. Floyd, J. Kikkawa, and V. Pattemore. 1975. Recent developments in the silviculture and management of sub-tropical rainforest in New South Wales. Proceedings of the Ecological Society of Australia 9:74-84.

Butler, B. J. A. R. L. C. 1998. Species richness, spatial variation, and abundance of the soil seed bank of a secondary tropical rain forest. Biotropica 30:214-222.

Caballe, G. 1984. The dynamics of woody liana communities in a rain forest of North- East Gabon. Terre Vie 39:3-36.

Cadenasso, M., and S. Pickett. 2001. Effect of edge structure on the flux of species into forest interiors. Conservation Biology 15:91-97.

Cadenasso, M. L., and S. T. A. Pickett. 2000. Linking forest edge structure to edge

165

function: mediation of herbivore damage. Journal of Ecology 88:31-44.

Cao, M., Y. Tang, C. Sheng, and J. Zhang. 2000. Viable seeds buried in the tropical forest soils of Xishuangbanna, SW China. Seed Science Research 10:255-264.

Carsten, L., F. Juola, T. Male, and S. Cherry. 2002. Host associations of lianas in a south-east Queensland rain forest. Journal of Tropical Ecology 18:107-120.

Cary, G. J. 2002. Importance of a changing climate for fire regimes in Australia. Pages 26-49 in R. Bradstock, J. Williams, and A. Gill, editors. Flammable Australia: Fire Regimes and Biodiversity of a Continent. Cambridge University Press, Melbourne.

Chazdon, R. L. 1988. Sunflecks and their importance to forest understory plants. Adv. Ecol. Res. 18:1-63.

Chazdon, R. L., and R. W. Pearcy. 1991. The importance of sunflecks for forest understory plants. BioScience 41:760-766.

Cheke, A., W. Nanakorn, and C. Yankoses. 1979. Dormancy and dispersal of seeds of secondary forest species under the canopy of a primary tropical rain forest in northern Thailand. Biotropica 11:88-95.

Chen, J., S. C. Saunders, T. R. Crow, R. J. Naiman, K. D. Brosofske, G. D. Mroz, B. L. Brookshire, and J. F. Franklin. 1999. Microclimate in forest ecosystem and landscape ecology. BioScience 49:288-297.

Chokkalingam, U., and W. De Jong. 2001. Secondary forest: a working definition and typology. International Forestry Review 3:19.

Church, R. J. 1997. Avian frugivory in a subtropical rainforest: eleven years of observations in Lamington National Park. The Sunbird 27:85-97.

Clark, D. B., and D. A. Clark. 1989. The role of physical damage in the seedling mortality regime of a neotropical rain forest. Oikos 55:225-230.

Clarke, F. G. 2002. The History of Australia. Greenwood Press, Westport, Connecticut. London.

Clarke, K. R., and M. Ainsworth. 1993. A method of linking multivariate community structure to environmental variables. Marine Ecology Progress Series 92:205- 219.

Clarke, K. R., and R. M. Warwick. 2001. Primer 5. PRIMER-E Ltd, University of Plymouth, Plymouth, UK.

Clifford, H. T. 1959. Seed dispersal by motor vehicles. The Journal of Ecology 47:311-315.

Cochrane, M., and W. Laurance. 2002. Fire as a large-scale edge effect in Amazonian

166

forests. Journal of Tropical Ecology 18:311-325.

Cofinas, M., and C. Creighton. 2001. Australian native vegetation assessment 2001. National Land and Water Resources Audit. Commonwealth of Australia, Canberra.

Colwell, R. K. 2005. EstimateS: Statistical estimation of species richness and shared

species from samples. Version 7.5. http://purlHTU .UTH oclc. org/estimates.U Department of Ecology & Evolutionary Biology, University of Connecticut Storrs, USA.

Connell, J., J. Tracey, and L. Webb. 1984. Compensatory recruitment, growth, and mortality as factors maintaining rain forest tree diversity. Ecological Monographs 54:141-164.

Connell, J. H. 1971. On the role of natural enemies in preventing competitive exclusion in some marine animals and in rain forest trees. Pages 298–312 in P. den Boer, and G. Gradwell, editors. Dynamics of numbers in populations. Center for Agricultural Publishing and Documentation, Osterbeek Wageningen, the Netherlands.

Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs. Science 199:1302.

Connell, J. H., and P. T. Green. 2000. Seedling dynamics over thirty-two years in a tropical rain forest tree. Ecology 81:568-584.

Cosgrove, R. 1996. Origin and development of Australian Aboriginal tropical rainforest culture: a reconsideration. Antiquity 70:900-912.

Crisp, M. D., S. Laffan, H. P. Linder, and A. Monro. 2001. Endemism in the Australian flora. Journal of Biogeography 28:183-198.

Cuaron, A. D. 2000. A global perspective on habitat disturbance and tropical rainforest mammals. Conservation Biology 14:1574-1579.

Cubina, A., and T. M. Aide 2001. The effect of distance from forest edge on seed rain and soil seed bank in a tropical pasture. Biotropica 33:260-267.

Curran, L., S. Trigg, A. Mcdonald, D. Astiani, Y. Hardiono, P. Siregar, I. Caniago, and E. Kasischke. 2004. Lowland forest loss in protected areas of Indonesian Borneo. Science 303:1000-1003.

Curran, L., and C. Webb. 2000. Experimental tests of the spatiotemporal scale of seed predation in mast-fruiting dipterocarpaceae. Ecological Monographs 70:129- 148.

Da Silva, J., C. Uhl, and G. Murray. 1996. Plant succession, landscape management, and the ecology of frugivorous birds in abandoned Amazonian pastures. Conservation Biology 10: 491-503.

167

Dalling, J., and S. Hubbell. 2002. Seed size, growth rate and gap microsite conditions as determinants of recruitment success for pioneer species. Journal of Ecology 90:557-568.

Dalling, J., S. Hubbell, and K. Silvera. 1998a. Seed dispersal, seedling establishment and gap partitioning among tropical pioneer trees. Journal of Ecology 86:674- 689.

Dalling, J., H. Muller-Landau, S. Wright, and S. Hubbell. 2002. Role of dispersal in the recruitment limitation of neotropical pioneer species. Journal of Ecology 90:714-727.

Dalling, J., M. Swaine, and N. Garwood. 1997. Soil seed bank community dynamics in seasonally moist lowland tropical forest, Panama. Journal of Tropical Ecology 13:659-680.

Dalling, J., M. Swaine, and N. Garwood. 1998b. Dispersal patterns and seed bank dynamics of pioneer trees in moist tropical forest. Ecology 79:564-578.

Dalling, J., K. Winter, and S. Hubbell. 2004. Variation in growth responses of neotropical pioneers to simulated forest gaps. Functional Ecology 18:725-736.

Delissio, L., and R. Primack. 2003. The impact of drought on the population dynamics of canopy-tree seedlings in an aseasonal Malaysian rain forest. Journal of Tropical Ecology 19:489-500.

Denslow, J. 1987. Tropical rainforest gaps and tree species diversity. Annual Review of Ecology and Systematics 18:431-451.

Denyer, K., B. Burns, and J. Ogden. 2006. Buffering of native forest edge microclimate by adjoining tree plantations. Austral Ecology 31:478-489.

Devlaeminck, R., B. Bossuyt, and M. Hermy. 2005. Inflow of seeds through the forest edge: evidence from seed bank and vegetation patterns. Plant Ecology 176:1- 17.

Dewalt, S., K. Ickes, R. Nilus, K. Harms, and D. Burslem. 2006. Liana habitat associations and community structure in a Bornean lowland tropical forest. Plant Ecology 186:203-216.

Dewalt, S. J., Stefan A. Schnitzer and Julie S. Denslow. 2000. Density and diversity of lianas along a chronosequence in a central Panamanian lowland forest. Journal of Tropical Ecology 16:1-19.

Didham, R., and J. Lawton. 1999. Edge structure determines the magnitude of changes in microclimate and vegetation structure in tropical forest fragments. Biotropica 31:17-30.

Didham, R. K. 1998. Altered leaf-litter decomposition rates in tropical forest fragments. Oecologia 116:397-406.

168

Didham, R. K., J. Ghazoul, N. E. Stork, and A. J. Davis. 1996. Insects in fragmented forests: A functional approach. Trends in Ecology & Evolution 11:255-260.

Dixon, K., S. Roche, and J. Pate. 1995. The promotive effect of smokederived from burnt native vegetation on seed germination of Western Australian plants. Oecologia 101:185-192

Douglas, I. 1977. Land use. Pages 21-24. The Border Ranges : A Land Use Conflict in Regional Perspective. the Royal Society of Queensland, Queensland.

Duff, G., and G. Stocker. 1989. The effects of frosts on rainforest/open forest ecotones in the highlands of North Queensland. Proc. Roy. Soc. Qld 100:49-54.

Edgington, E. S. 1980. Randomization Tests. Marcel Decker, New York, USA.

Enright, N., and A. Kintrup. 2001. Effects of smoke, heat and charred wood on the germination of dormant soil-stored seeds from a Eucalyptus baxteri heathy- woodland in Victoria, SE Australia. Austral Ecology 26:132-141.

Enright, N. J., D. Goldblum, P. Ata, and D. H. Ashton. 1997. The independent effects of heat, smoke and ash on emergence of seedlings from the soil seed bank of heathy Eucalyptus woodland in Grampians (Gariwerd) National Park, western Victoria. Australian Journal of Ecology 22:81-88.

Enright, N. J., R. Marsula, B. B. Lamont, and C. Wissel. 1998. The ecological significance of canopy seed storage in fire-prone environments: a model for non-sprouting shrubs. Journal of Ecology 86:946-959.

Eriksson, O., and J. Ehrlén. 1992. Seed and microsite limitation of recruitment in plant populations. Oecologia 91:360-364.

Erskine, P. 2002. Land clearing and forest rehabilitation in the Wet Tropics of north Queensland, Australia. Ecological Management and Restoration 3:135-136.

Ewel, J., C. Berish, B. Brown, N. Price, and J. Raich. 1981. Slash and burn impacts on a Costa Rican wet forest site. Ecology 62:816-829.

Fagan, C., C. A. Peres, and J. Terborgh. 2006. Tropical forests: a protected-area strategy for the twenty-first century. Pagers 415-432 in W. Laurance, and C. Peres, editors. Emerging threats to tropical forests. University of Chicago Press, Chicago.

Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology, Evolution and Systematics 34:487-515.

FAO 2001. The Global Forest Resources Assessment 2000 Page 479. Main Report. FAO Forestry Paper 140. FAO, Rome.

Fensham, R. J. 1997. Aboriginal fire regimes in Queensland, Australia: analysis of the explorer's record. Journal of Biogeography 24:11-22.

169

Ferguson, R. N., and D. R. Drake. 1999. Influence of vegetation structure on spatial patterns of seed deposition by birds. New Zealand Journal of Botany 37:671- 677.

Ferreira, L., and W. Laurance. 1997. Effects of forest fragmentation on mortality and damage of selected trees in central Amazonia. Conservation Biology 11:797- 801.

Floyd, A. G. 1976. Effect of burning on regeneration from seeds in wet sclerophyll forest. Australian Forestry 39:210-220.

Floyd, A. G. 1987. Status of rainforest in northern New South Wales. Pages 95-117. The rainforest legacy : Australian national rainforests study. Australian Govt. Pub. Service, 1987-1991, Canberra, A.C.T. .

Floyd, A. G. 1990. Australian Rainforests in New South Wales. Surrey Beatty and Sons, Sydney.

Forman, R., and L. Alexander. 1998. Roads and their major ecological effects. Annual Review of Ecology and Systematics 29:207-231.

Foster, S., and C. H. Janson. 1985. The relationship between seed size and establishment conditions in tropical woody plants. Ecology 66:773-780.

Fox, B., J. Ttaylor, M. Fox, and C. Williams. 1997. Vegetation changes across edges of rainforest remnants. Biological conservation 82:1-13.

Fox, J., D. Truong, A. T. Rambo, N. P. Tuyen, L. E. T. Cuc, and S. Leisz. 2000. Shifting cultivation: a new old paradigm for managing tropical forests. BioScience 50:521-528.

Freebody, K. 2007. Rainforest revegetation in the uplands of the Australian Wet Tropics: The Eacham Shire experience with planting models, outcomes and monitoring issues. Ecological Management and Restoration 8:140-143.

Frost, W. 2004. Tourism, rainforests and worthless lands: the origins of national parks in Queensland. Tourism Geographies 6:493-507.

Galindo-Gonzalez, J., S. Guevara, and V. J. Sosa. 2000. Bat- andbird-generated seed rains at Isolated trees in pastures in a tropical rainforest. Conservation Biology 14:1693-1703.

Garcia, L. V. 2004. Escaping the Bonferroni iron claw in ecological studies. Oikos 105:657-663.

Garwood, N. C. 1989. Tropical soil seed banks: a review. Pages 149-209 in M. A. Leck, V. T. Parker, and R. L. Simpson, editors. Ecology of Soil Seed Banks. Academic Press, San Diego

Garwood, N. C. 1996. Functional morphology of tropical tree seedlings. Pages 59-117

170

in T. Whitmore, and M. Swaine, editors. Ecology of Tropical Forest Tree Seedlings. UNESCO and Parthenon Publishing, Paris.

Gascon, C., T. Lovejoy, R. Bierregaard Jr, J. Malcolm, P. Stouffer, H. Vasconcelos, W. Laurance, B. Zimmerman, M. Tocher, and S. Borges. 1999. Matrix habitat and species richness in tropical forest remnants. Biological Conservation 91:223- 229.

Gascon, C., G. B. Williamson, and G. A. B. Da Fonseca. 2000. Receding forest edges and vanishing reserves. Science 288:1356-1358.

Gehring, C., S. Park, and M. Denich. 2004. Liana allometric biomass equations for Amazonian primary and secondary forest. Forest Ecology and Management 195:69-83.

Gelbard, J., and J. Belnap. 2003. Roads as conduits for exotic plant invasions in a semiarid landscape. The Journal of the Society for Conservation Biology 17:420-432.

Gentle, C. B., and J. A. Duggin. 1997. Lantana camara L. invasions in dry rainforest - open forest ecotones: the role of disturbances associated with fire and cattle grazing. Austral Ecology 22:298-306.

Gentry, A. 1991. The distribution and evolution of climbing plants. Pages 3–49 in F. Putz, and H. Mooney, editors. The Biology of Vines. Cambridge University Press, Cambridge.

Gerwing, J. 2001. Testing liana cutting and controlled burning as silvicultural treatments for a logged forest in the eastern Amazon. Journal of Applied Ecology 38:1264-1276.

Gerwing, J., and C. Uhl. 2002. Pre-logging liana cutting reduces liana regeneration in logging gaps in the eastern Brazilian Amazon. Ecological Applications 12:1642-1651.

Gerwing, J. J., and D. L. Farias. 2000. Integrating liana abundance and forest stature into an estimate of total aboveground biomass for an eastern Amazonian forest. Journal of Tropical Ecology 16:327-335.

Gerwing, J. J., S. A. Schnitzer, R. J. Burnham, F. Bongers, J. Chave, S. J. Dewalt, C. E. N. Ewango, R. Foster, D. Kenfack, M. Martinez-Ramos, M. Parren, N. Parthasarathy, D. R. Perez-Salicrup, F. E. Putz, and D. W. Thomas. 2006. A standard protocol for liana censuses. Biotropica 38:256-261.

Gilbert, B., J. Wright, H. Muller-Landau, K. Kitajima, and A. Hernandéz. 2006. Life history trade-offs in tropical trees and lianas. Ecology 87:1281-1288.

Gilbert, G. S., K. E. Harms, D. N. Hamill, and S. P. Hubbell. 2001. Effects of seedling size, El Nino drought, seedling density, and distance to nearest conspecific adult on 6-year survival of Ocotea whitei seedlings in Panama. Oecologia

171

127:509-516.

Gill, A. 1981. Adaptive responses of Australian vascular plant spccies to fores. Pages 243-272 in A. Gill, R. Groves, and I. Noble, editors. Fire and the Australian Biota. Australian Academy of Science, Canberra.

Glanznig, A. 1995. Native Vegetation Clearance, Habitat Loss and Biodiversity Decline: An overview of recent native vegetation clearance in Australia and its implications for biodiversity. Biodiversity Series, Paper No. 6. Australian Government Department of the Environment and Water Resources.

Gomez-Pompa, A., C. Vazquez-Yanes, and S. Guevara. 1972. The tropical rain forest: a nonrenewable resource. Science 177:762-765.

Goosem, M. 1997. Internal fragmentation: the effects of roads, highways, and powerline clearings on movements and mortality of rainforest vertebrates. Pages 241–255 in L. WF, editor. Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities. University of Chicago Press, Chicago.

Goosem, M., and S. M. Turton. 2006. Weed Incursions Along Roads and Powerlines in the Wet Tropics World Heritage Area: The Potential of Remote Sensing as an Indicator of Weed Infestations. Cooperative Reseaerch Center for Tropical Rainforest Ecology and Management. Rainforest CRC, Cairns (199 pp).

Graham, A., and M. Hopkins. 1990. Soil seed banks of adjacent unlogged rainforest types in north Queensland. Australian Journal of Botany 38:261-268.

Graham, A., J. G. Tracy, and M. S. Hopkins. 1977. The forest. Pages 1-6. The Border Ranges : A Land Use Conflict in Regional Perspective. the Royal Society of Queensland, Queensland.

Green, P. T., and P. A. Juniper. 2004. Seed-seedling allometry in tropical rain forest trees: seed mass-related patterns of resource allocation and the'reserve effect'. Journal of Ecology 92:397-408.

Grubb, P. 1996. Rainforest dynamics: the need for new paradigms. Tropical Rainforest Research 74:215–233.

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.

Grubb, P. J., and D. J. Metcalfe. 1996. Adaptation and Inertia in the Australian Tropical Lowland Rain-Forest Flora: Contradictory Trends in Intergeneric and Intrageneric Comparisons of Seed Size in Relation to Light Demand. Functional Ecology 10:512-520.

Guevara, S., S. E. Purata, and E. Maarel. 1986. The role of remnant forest trees in tropical secondary succession. Plant Ecology 66:77-84.

172

Hall, J. 1986. Exploratory excavation at Bushrangers Cave (site LA: A11), a 6,000 year old campsite in southeast Queensland: preliminary results. Australian Archaeology 22:88-103.

Harden, G. 1991. Flora of New South Wales. New South Wales University Press, Sydney.

Hardy, O., and X. Vekemans. AutocorQ 2.0 - a spatial autocorrelation analysis program for quantitative data. http://www.ulb.ac.be/sciences/ecoevol/autocorq.html

Harms, K. E., S. J. Wright, O. Calderon, A. Hernandez, and E. A. Herre. 2000. Pervasive density-dependent recruitment enhances seedling diversity in a tropical forest. Nature 404:493-495.

Harper, J. L. 1977. Population Biology of Plants. Academic Press, New York.

Harper, K. A., S. E. Macdonald, P. J. Burton, J. Chen, K. D. Brosofske, S. C. Saunders, E. S. Euskirchen, D. A. R. Roberts, M. S. Jaiteh, and P.-A. Esseen. 2005. Edge influence on forest structure and composition in fragmented landscapes. Conservation Biology 19:768-782.

Harrington, G. N., and K. D. Sanderson. 1994. Recent contraction of wet sclerophyll forest in the wet tropics of Queensland due to invasion by rainforest. Pacific Conservation Biology 1:319-327.

Hegarty, E., and G. Caballé. 1991. Distribution and abundance of vines in forest communities. Pages 313–335 in F. Putz, and H. Mooney, editors. The Biology of Vines. Cambridge University Press, Cambridge.

Hegarty, E. E. 1989. Canopy dynamics of lianes and trees in subtropical rainforest. Austral Ecology 14:559-560.

Hester, A. J., and R. J. Hobbs. 1992. Influence of fire and soil nutrients on native and non-native annuals at remnant vegetation edges in the western Australian Wheatbelt. Journal of Vegetation Science 3:101-108.

Hilbert, D. W., B. Ostendorf, and M. S. Hopkins. 2001. Sensitivity of tropical forests to climate change in the humid tropics of north Queensland. Austral Ecology 26:590-603.

Hill, J., and P. Curran. 2001. Species composition in fragmented forests: conservation implications of changing forest area. Applied Geography 21:157-174.

Hill, R., P. Griggs, and B. B. N. Incorporated. 2000. Rainforests, agriculture and aboriginal fire-regimes in wet tropical Queensland, Australia. Australian Geographical Studies 38:138-157.

Hobbs, R. J., and A. J. M. Hopkins. 1990. From frontier to fragments: European impact on Australia’s vegetation. Proceedings of the Ecological Society of

173

Australia 16:93–114.

Holl, K., M. Loik, E. Lin, and I. Samuels. 2000. Tropical montane forest restoration in Costa Rica: overcoming barriers to dispersal and establishment. Restoration Ecology 8:339-349.

Holl, K. D. 1998. Do bird perching structures elevate seed rain and seedling establishment in abandoned tropical pasture? Restoration Ecology 6:253-261.

Holl, K. D. 1999. Factors limiting tropical rain forest regeneration in abandoned pasture: seed rain, seed germination, microclimate, and soil. Biotropica 31:229-242.

Hood, G. 2004. Pop tools version 2.6. Excel Add-In. Pest Animal Control CRC, CSIRO, Canberra, Australia.

Hooper, E., P. Legendre, and R. Condit. 2005. Barriers to forest regeneration of deforested and abandoned land in Panama. Journal of Applied Ecology 42:1165-1174.

Hopkins, M., and A. Graham. 1983. The species composition of soil seed banks beneath lowland tropical rainforests in North Queensland, Australia. Biotropica 15:90-99.

Hopkins, M., and A. Graham. 1984. Viable soil seed banks in disturbed lowland tropical rainforest sites in North Queensland. Australian Journal of Ecology 9:71-79.

Hopkins, M., and A. Graham. 1987. The viability of seeds of rainforest species after experimental soil burials under tropical wet lowland forest in north-eastern Australia. Austral Ecology 12:97-108.

Hopkins, M., J. Tracey, and A. Graham. 1990a. The size and composition of soil seed- banks in remnant patches of three structural rainforest types in North Queensland. Australian Journal of Ecology 15:43-50.

Hopkins, M. S. 1975. Species patterns and diversity in the subtropical rain forest Thesis (Ph.D., Page 251). Department of Botany. University of Queensland, Brisbane.

Hopkins, M. S., J. Ash, A. W. Graham, J. Head, and R. K. Hewett. 1993. Charcoal evidence of the spatial extent of the eucalyptus woodland expansions and rainforest contractions in North Queensland during the late Pleistocene. Journal of Biogeography 20:357-372.

Hopkins, M. S., A. W. Graham, R. Hewett, J. Ash, and J. Head. 1990b. Evidence of late Pleistocene fires and eucalypt forest from a north Queendland humid tropical Rainforest site. Australian Journal of Ecology 15:345-347.

Hopkins, M. S., J. Head, J. E. Ash, R. K. Hewett, and A. W. Graham. 1996. Evidence

174

of a Holocene and continuing recent expansion of lowland rain forest in humid, tropical North Queensland. Journal of Biogeography 23:737-745.

Hopkins, M. S., J. Kikkawa, A. Graham, W,, J. G. Tracy, and L. J. Webb. 1977. An ecological basis for management of rain forest. Pages 1-6. The Border Ranges: A Land Use Conflict in Regional Perspective. the Royal Society of Queensland, Queensland.

Horne, R., and J. Hickey. 1991. Ecological sensitivity of Australian rainforests to selective logging. Austral Ecology 16:119-129.

Hubbell, S., R. Foster, S. O’brien, K. Harms, R. Condit, B. Wechsler, S. Wright, and S. De Lao. 1999. Light-gap disturbances, recruitment limitation, and tree diversity in a neotropical forest. Science 283:554-557.

Hughes, L., and M. Westoby. 1992a. Effect of diaspore characteristics on removal of seeds adapted for dispersal by ants. Ecology 73:1300-1312.

Hughes, L., and M. Westoby. 1992b. Fate of seeds adapted for dispersal by ants in Australian sclerophyll vegetation. Ecology 73:1285-1299.

Ibarra-Manriquez, G., and M. Martinez-Ramos. 2002. Landscape variation of liana communities in a Neotropical rain forest. Plant Ecology 160:91-112.

Jackson, R., and G. Stephenson. 1986. Lamington National Park Road. The Road Builders 1:31-33.

Jackson, W. 1978. Ecological Drift: an argument against the continued practise of hazard reduction burning. Pages 98-101 in H. Gee, and F. J, editors. The southwest bBook: a Tasmanian Wilderness. Australian Conservation Foundation, Melbourne.

James, A. N., K. J. Gaston, and A. Balmford. 1999. Balancing the Earth’s accounts. Nature 401:323-324.

Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. The American Naturalist 104:501-528.

Janzen, D. H. 1974. Tropical blackwater rivers, animals, and mast fruiting by the dipterocarpaceae. Biotropica 6:69-103.

Janzen, D. H. 1983. No park Is an island - increase in interference from outside as park size decreases. Oikos 41:402-410.

Jarrott, J. K. 1977. The history. Pages 1-6. The Border Ranges : A Land Use Conflict in Regional Perspective. The Royal Society of Queensland, Queensland.

Jongman, R. H. G., C. J. F. Ter Braak, and O. F. R. Van Tongeren 1995. Data analysis in community and landscape ecology. Cambridge University Press Cambridge, United Kingdom.

175

Kanowski, J., and C. P. Catterall 2007. Monitoring Revegetation Projects for Biodiversity in Rainforest Landscapes. Tookit Version 1, Revision 1. Marine and Tropical Sciences Research Facility, Cairns.

Kanowski, J., C. P. Catterall, G. W. Wardell-Johnson, H. Proctor, and T. Reis. 2003. Development of forest structure on cleared rainforest land in eastern Australia under different styles of reforestation. Forest Ecology and Management 183:265-280.

Kapos, V. 1989. Effects of isolation on the water status of forest patches in the Brazilian Amazon. Journal of Tropical Ecology 5:173-185.

Kapos, V., E. Wandelli, J. L. Camargo, and G. Ganade. 1997. Edge-related changes in environment and plant responses due to forest fragmentation in central Amazonia. Pages 33–44 in W. F. Laurance, and R. O. Bierregaard, Jr., editors. Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities. University of Chicago Press, Chicago.

Keeley, J. E., and W. J. Bond. 1999. Mast flowering and semelparity in bamboos: the bamboo fire cycle hypothesis. The American Naturalist 154:383-391.

Kellman, M., R. Tackaberry, and J. Meave. 1996. The consequences of prolonged fragmentation: lessons from tropical gallery forests. Pages 19-36 in J. Schelhas, and R. Greenberg, editors. Forest Patches in Tropical Landscapes. Island Press, Washington, DC, USA.

Kelly, D., and V. L. Sork. 2002. Mast seeding in perennial plants: why, how, where? Annual Review of Ecology and Systematics 33:427-447.

Kershaw, A. P. 1978. Record of last interglacial–glacial cycle from northeastern Queensland. Nature 272:159-161.

Kershaw, A. P. 1981. Quaternary vegetation and environments. Pages 81–101 in A. Keast, editor. Ecological Biogeography of Australia. Dr W. Junk b.v. Publishers, London.

Kershaw, A. P. 1986. Climatic change and Aboriginal burning in north-east Australia during the last two glacial/interglacial cycles. Nature 322:47-48.

Kershaw, A. P. 1992. The development of rainforest-savanna boundaries in tropical Australia. Pages 255–273 in P. A. Furley, J. Proctor, and J. A. Ratter, editors. Nature and Dynamics of Forest-Savanna Boundaries. Chapman and Hall, London.

Kershaw, A. P. 1994. Pleistocene vegetation of the humid tropics of northeastern Queensland, Australia. Palaeogeography, Palaeoclimatology, Palaeoecology 109:399-412.

Kershaw, A. P., J. S. Clark, A. M. Gill, and D. M. D’costa. 2002. A history of fire in Australia. Pages 1-25 in R. A. Bradstock, J. E. Williams, and A. M. Gill,

176

editors. Flammable Australia: the Fire Regimes and Biodiversity of a Continent. Cambridge University Press.

Kitajima, K. 1994. Relative importance of photosynthetic traits and allocation patterns as correlates of seedling shade tolerance of 13 tropical trees. Oecologia 98:419-428.

Kitching, R. L., S. L. Boulter, G. Vickerman, M. Laidlaw, K. Hurley, and P. S. Grimbacher 2005. The Comparative Assessment of Arthropod and Tree Biodiversity in Old-world Rainforests. The Rainforest CRC / Earthwatch Protocol Manual (2nd ed.). Cooperative Research Centre for Tropical Rainforest Ecology and Management, Cairns.

Laidlaw, M., M. Olsen, R. Kitching, and M. Greenway. 2000. Tree floristic and structural characteristics of one hectare of subtropical rainforest in Lamington National Park, Queensland. Proceedings of the Royal Society of Queensland 109:91-105.

Lamb, D., P. Erskine, and J. Parrotta. 2005. Restoration of degraded tropical forest landscapes. Science 310:1628-1632.

Lamont, B. B., D. C. Le Maitre, R. M. Cowling, and N. J. Enright. 1991. Canopy seed storage in woody plants. The Botanical Review 57:277-317.

Laurance, W. 1991a. Edge effects in tropical forest fragments: Application of a model for the design of nature reserves. Biological Conservation 57:205-219.

Laurance, W. 2000. Do edge effects occur over large spatial scales? Trends in Ecology & Evolution 15:134-135.

Laurance, W. 2004a. Forest-climate interactions in fragmented tropical landscapes. Philosophical Transactions: Biological Sciences 359:345-352.

Laurance, W. 2004b. The perils of payoff: corruption as a threat to global biodiversity. Trends in Ecology & Evolution 19:399–401.

Laurance, W., P. Delamonica, S. Laurance, H. Vasconcelos, and T. Lovejoy. 2000. Rainforest fragmentation kills big trees. Nature 404:836.

Laurance, W., L. Ferreira, J. Merona, and S. Laurance. 1998a. Rain forest fragmentation and the dynamics of Amazonian tree communities. Ecology 79:2032-2040.

Laurance, W., L. Ferreira, J. Merona, S. Laurance, R. Hutchings, and T. Lovejoy. 1998b. Effects of forest fragmentation on recruitment patterns in Amazonian tree communities. Conservation Biology 12:460-464.

Laurance, W., T. Lovejoy, H. Vasconcelos, E. Bruna, R. Didham, P. Stouffer, C. Gascon, R. Bierregaard, S. Laurance, and E. Sampaio. 2002. Ecosystem decay of Amazonian forest fragments: a 22-year investigation. Conservation Biology

177

16:605-618.

Laurance, W., G. Williamson, P. Delamonica, A. Oliveira, T. Lovejoy, C. Gascon, and L. Pohl. 2001a. Effects of a strong drought on Amazonian forest fragments and edges. Journal of Tropical Ecology 17:771-785.

Laurance, W. F. 1991b. Edge effects in tropical forest fragments: Application of a model for the design of nature reserves. Biological Conservation 57:205-219.

Laurance, W. F. 1997. Hyper-disturbance parks: edge effects and isolated rainforest reserves in tropical Australia. Pages 71-83 in W. F. Laurance, and R. O. Bierregaard, Jr., editors. Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities. University of Chicago Press, Chicago.

Laurance, W. F. 2007. Have we overstated the tropical biodiversity crisis? Trends in Ecology & Evolution 22:65-70.

Laurance, W. F., and G. N. Harrington. 1997. Ecological associations of feeding sites of feral pigs in the Queensland wet tropics. Wildlife Research 24:579-590.

Laurance, W. F., S. G. Laurance, L. V. Ferreira, J. M. Rankin-De Merona, C. Gascon, and T. E. Lovejoy. 1997. Biomass collapse in Amazonian forest fragments. Science 278:1117.

Laurance, W. F., S. G. Laurance, and D. P. 1998c. Tropical forest fragmentation and greenhouse gas emissions. Forest Ecology and Management 110:173-180.

Laurance, W. F., H. E. M. Nascimento, S. G. Laurance, A. Andrade, J. E. L. S. Ribeiro, J. P. Giraldo, T. E. Lovejoy, R. Condit, J. Chave, K. E. Harms, and S. D'angelo. 2006a. Rapid decay of tree-community composition in Amazonian forest fragments. PNAS 103:19010-19014.

Laurance, W. F., H. E. M. Nascimento, S. G. Laurance, A. C. Andrade, P. M. Fearnside, J. E. L. Ribeiro, and R. L. Capretz. 2006b. Rain forest fragmentation and the proliferation of successional trees. Ecology 87:469-482.

Laurance, W. F., and C. A. Peres, editors. 2006. Emerging Threats to Tropical Forests. University of Chicago press, Chicago.

Laurance, W. F., D. Perez-Salicrup, P. Delamonica, P. M. Fearnside, S. D'angelo, A. Jerozolinski, L. Pohl, and T. E. Lovejoy. 2001b. Rain forest fragmentation and the structure of Amazonian liana communities. Ecology 82:105-116.

Lawes, M. J., B. C. C. Lamb, and S. Boudreau. 2005. Area-but no edge-effect on woody seedling abundance and species richness in old Afromontane forest fragments. Journal of Vegetation Science 16:363-372.

Leakey, A. D. B., J. D. Scholes, and M. C. Press. 2005. Physiological and ecological significance of sunflecks for dipterocarp seedlings. Journal of Experimental

178

Botany 56:469-482.

Lesley, H. 1989. Prehistoric Aboriginal impacts on Australian vegetation: An assessment of the evidence. Australian Geographer 20:37-46.

Levey, D. J. B., Margaret M. 1993. Complex ant-plant interactions ants as secondary dispersers. Ecology 74:1802-1812.

Lewis, S. 2006. Tropical forests and the changing earth system. Philosophical Transactions of the Royal Society B: Biological Sciences 361:195-210.

Lewis, S., Y. Malhi, and O. Phillips. 2004. Fingerprinting the impacts of global change on tropical forests. Philosophical Transactions: Biological Sciences 359:437-462.

Lewis, S. L., Tanner, Edmund V. J. 2000. Effects of above-and blowground competition on growth and survival of rain forest tree seedlings. Ecology 81:2525–2538.

Lin, L., M. Cao, Y. He, J. M. Baskin, and C. C. Baskin. 2006. Nonconstituent species in soil seed banks as indicators of anthropogenic disturbance in forest fragments. Canadian Journal of Forestry Research 36:2300-2316.

Lovejoy, T. E., R. O. Bierregaard, J. M. Rankin, and H. O. R. Schubart. 1984. Ecological dynamics of tropical forest fragments. Pages 377-384 in S. L. Sutton, T. C. Whitmore, and A. C. Chadwick, editors. Tropica rain forest: ecology and management. Blackwell Scientific, Oxford.

Lusk, C., and C. Kelly. 2003. Interspecific variation in seed size and safe sites in a temperate rain forest. New Phytologist 158:535-541.

Lusk, C. H. 1999. Long-lived light-demanding emergents in southern temperate forests: the case of Weinmannia trichosperma (Cunoniaceae) in Chile. Plant Ecology 140:111-115.

Macdougall, A., and M. Kellman. 1992. The understorey light regime and patterns of tree seedlings in tropical riparin forest patches. Journal of Biogeography 19:667-675.

Magurran, A. E. 2004. Measuring Biological Diversity. Blackwell Publishing.

Malizia, A., and H. R. Grau. 2006. Liana-host tree associations in a subtropical montane forest of north-western Argentina. Journal of Tropical Ecology 22:331-340.

Margules, C. R., and R. L. Pressey. 2000. Systematic conservation planning. Nature 405:243-253.

Marod, D., U. Kutintara, H. Tanaka, and T. Nakashizuka. 2002. The effects of drought and fire on seed and seedling dynamics in a tropical seasonal forest in

179

Thailand. Plant Ecology 161:41-57.

Martinez-Garza, C., and H. F. Howe. 2003. Restoring tropical diversity: beating the time tax on species loss. Journal of Applied Ecology 40:423-429.

Mcdonald, W., and M. Thomas 1990. The Flora of Lamington National Park: A Preliminary Checklist of the Vascular Plant Species. Queensland Department of Primary Industries, Brisbane.

McDonald, W., and M. B. Thomas. 1989. Rare and Threatened Plants of Queensland (2nd ed.). Queendland Department of Primary Industries, Queensland Government Printers, Brisbane.

McDonald, W. J. F., and W. G. Whiteman. 1979. Moreton Region Vegetation Map Series: Explanatory Booklet for Murwillumbah Sheet. Botany Branch, Queendland Department of Primary Industarles, Queensland Government, Brisbane.

Mesquita, R., P. Delamonica, and W. Laurance. 1999. Effect of surrounding vegetation on edge-related tree mortality in Amazonian forest fragments-a ten- year study. Biological Conservation 91:129-134.

Metcalfe, D. J., and I. M. Turner. 1998. Soil seed bank from lowland rain forest in Singapore: canopy-gap and litter-gap demanders. Journal of Tropical Ecology 14:103-108.

Miller, P. 2000. Effects of deforestation on seed banks in a tropical deciduous forest of western Mexico. Journal of Tropical Ecology 15:179-188.

Molino, J., and D. Sabatier. 2001. Tree diversity in tropical rain forests: a validation of the intermediate disturbance hypothesis. Science 294:1702-1704.

Montgomery, R. A. 2004. Effects of understory foliage on patterns of light attenuation near the forest floor. Biotropica 36:33-39.

Montgomery, R. A., and R. L. Chazdon. 2001. Forest structure, canopy architecture, and light transmittance in tropical wet forests. Ecology 82:2707-2718.

Moran, M. D. 2003. Arguments for rejecting the sequential Bonferroni in ecological studies. Oikos 100:403-405.

Murcia, C. 1995. Edge effects in fragmented forests - implications for conservation. Trends in Ecology & Evolution 10:58-62.

Murren, C. J. 2002. Effects of habitat fragmentation on pollination: pollinators, pollinia viability and reproductive success. Journal of Ecology 90:100-107.

Myers, N. 1993. Tropical forests: The main deforestation fronts. Environmental Conservation 20:9-16.

180

Myers, N. 1994. Tropical deforestation: rates and patterns. Pages 27-40 in K. Brown, and D. Pearce, editors. The Causes of Tropical Deforestation. UCL Press Limited, London.

Naaf, T., and M. Wulf. 2007. Effects of gap size, light and herbivory on the herb layer vegetation in European beech forest gaps. Forest Ecology and Management 244:141-149.

Nadolny, C. 1999. Recruitment and survival of Argyrodendron actinophyllum seedlings in an Australian rainforest. Australian Journal of Ecology 24:258.

Nakagawa, M., K. Tanaka, T. Nakashizuka, T. Ohkubo, T. Kato, T. Maeda, K. Sato, H. Miguchi, H. Nagamasu, and K. Ogino. 2000. Impact of severe drought associated with the 1997–1998 El Niño in a tropical forest in Sarawak. Journal of Tropical Ecology 16:355-367.

Nascimento, H. E. M., A. Andrade, J. L. C. Camargo, W. F. Laurance, S. G. Laurance, and J. E. L. Ribeiro. 2006. Effects of the surrounding matrix on tree recruitment in Amazonian forest fragments. Conservation Biology 20:853-860.

Nepstad, D., A. Verissimo, A. Alencar, C. Nobre, E. Lima, P. Lefebvre, P. Schlesinger, C. Potter, P. Moutinho, and E. Mendoza. 1999. Large-scale impoverishment of Amazonian forests by logging and fire. Nature 398:505±508.

Nepstad, D. C., C. Uhl, C. A. Pereira, and J. M. C. Da Silva. 1996. A comparative study of tree establishment in abandoned pasture and mature forest of eastern Amazonia. Oikos 76:25-39.

Nicotra, A. B., R. L. Chazdon, and S. V. B. Iriarte. 1999. Spatial heterogeneity of light and woody seedling regeneration in tropical wet forests. Ecology 80:1908- 1926.

O'reilly, B. 1944. Green Mountains. W.R.Smith and Paterson, Brisbane.

Ogilvie, P. 2006. A brief park history in L. Hutley, editor. A Guide to Lamington National Park. Envirobook, Canterbury, NSW, Australian.

Oliveira-Filho, A. T., J. Marcio De Mello, and J. R. S. Scolforo. 1997. Effects of past disturbance and edges on tree community structure and dynamics within a fragment of tropical semideciduous forest in south-eastern Brazil over a five- year period (1987-1992). Plant Ecology 131:45-66.

Olsen, M. 1990. Aspects of the Ecology and Physiology of Rainforest Systems Following Large Scale Natural Disturbance.Thesis ( Ph.D., Page 242). School of Integrative Biology. University of Queensland

Olsen, M., and D. Lamb. 1988. Recovery of subtropical rainforest following storm damage. Proc. Ecol. Soc. Aust 15:297–301.

Olupot, W., and C. A. Chapman. 2006. Human encroachment and vegetation change

181

in isolated forest reserves: the case of Bwindi Impenetrable National Park, Uganda. Pages 127-148 in W. F. Laurance, and C. A. Peres, editors. Emerging Threats to Tropical Forests. The University of Chicago Press, Chicago and London.

Opdam, P., and D. Wascher. 2004. Climate change meets habitat fragmentation: linking landscape and biogeographical scale levels in research and conservation. Biological Conservation 117:285-297.

Osunkoya, O. O. 1996. Light requirements for regeneration in tropical forest plants: Taxon-level and ecological attribute effects. Austral Ecology 21:429-441.

Otero-Arnaiz, A., S. Castillo, J. Weave, and G. Ibarra-Manriquez. 1999. Isolated pasture trees and the vegetation under their canopies in the chiapas Coastal Plain, Mexico. Biotropica 31:243-254.

Parendes, L. A., and J. A. Jones. 2000. Role of light availability and dispersal in exotic plant invasion along roads and streams in the H. J. Andrews Experimental Forest, Oregon. Conservation Biology 14:64-75.

Pavlov, P. M., J. Hone, and L. A. Moore. 1992. Feral pigs, rainforest conservation and exotic disease in north Queensland. Wildlife Research 19:179-193.

Perez-Salicrup, D. R. 2001. Cost and efficiency of cutting lianas in a lowland liana forest of Bolivia. Biotropica 33:324-329.

Perez-Salicrup, D. R., V. L. Sork, and F. E. Putz. 2001. Lianas and trees in a liana forest of Amazonian Bolivia. Biotropica 33:34-47.

Peters, H. A. 2001. Clidemia hirta invasion at the pasoh forest reserve: An unexpected plant invasion in an undisturbed tropical forest. Biotropica 33:60-68.

Phillips, O., R. Martínez, A. Mendoza, T. Baker, and P. Vargas. 2005. Large lianas as hyperdynamic elements of the tropical forest canopy. Ecology 86:1250-1258.

Phillips, O. L., R. V. Martinez, L. Arroyo, T. R. Baker, T. Killeen, S. L. Lewis, Y. Malhi, A. M. Mendoza, D. Neill, P. N. Vargas, M. Alexiades, C. Ceron, A. Di Fiore, T. Erwin, A. Jardim, W. Palacios, M. Saldias, and B. Vinceti. 2002. Increasing dominance of large lianas in Amazonian forests. Nature 418:770- 774.

Phillips, O. L., S. A. M. Rose, A. M. Mendoza, and P. N. Vargas. 2006. Resilience of southwestern Amazon forests to anthropogenic edge effects. Conservation Biology 20:1698-1710.

Pimm, S. L., and T. Brooks. 1999. The sixth extinction: How large, where, and when. Pages 46–62 in P. H. Raven, and T. Williams, editors. Nature and Human Society: The Quest for a Sustainable World. National Academy, Washinton.

Pohlman, C. L., S. M. Turton, and M. Goosem. 2007. Edge effects of linear canopy

182

openings on tropical rain forest understory microclimate. Biotropica 39:62-71.

Popma, J., and F. Bongers. 1988. The effect of canopy gaps on growth and morphology of seedlings of rain forest species. Oecologia 75:625-632.

Putz, F. 1983. Treefall pits and mounds, buried seeds, and the importance of soil disturbance to pioneer trees on Barro Colorado Island, Panama. Ecology 64:1069-1074.

Putz, F. 1984a. How trees avoid and shed lianas. Biotropica 16:19-23.

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

Putz, F. 2004. Vine Ecology. www.ecology.info/vines.htmHTU .UTH

Putz, F., and S. Appanah. 1987. Buried seeds, newly dispersed seeds, and the dynamics of a lowland forest in Malaysia. Biotropica 19:326-333.

Putz, F., and P. Chai. 1987. Ecological studies of lianas in Lambir National park, Sarawak, Malaysia. Journal of Ecology 75:523-531.

Putz, F., and H. Mooney 1991. The Biology of Vines. Cambridge, UK.

Putz, F. E., and N. M. Holbrook. 1991. Biomechanical studies of vines. Pages 73–97. The biology of vines. Cambridge University Press, Cambridge.

Quinn, G. P., and M. J. Keough 2002. Experimental Design and Data Analysis for Biologists. Blackwell Synergy.

Read, T., S. Bellairs, D. Mulligan, and D. Lamb. 2000. Smoke and heat effects on soil seed bank germination for the re-establishment of a native forest community in New South Wales. Austral Ecology 25:48-57.

Restrepo, C., and A. Vargas. 1999. Seeds and seedlings of two neotropical montane understory shrubs respond differently to anthropogenic edges and treefall gaps. Oecologia 119:419-426.

Rice, W. R. 1989. Analyzing Tables of Statistical Tests. Evolution 43:223-225.

Richards, P. W. 1996. The Tropical Rain Forest. Cambridge University Press, Cambridge.

Roberts, R. G., T. F. Flannery, L. K. Ayliffe, H. Yoshida, J. M. Olley, G. J. Prideaux, G. M. Laslett, A. Baynes, M. A. Smith, and R. Jones. 2001. New ages for the last Australian megafauna: Continent-wide extinction about 46,000 years ago. Science 292:1888-1892.

Russell-Smith, J., and P. Stanton. 2002. Fire regimes and fire management of rainforest communities across northern Australia. Pages 329–350 in R. A.

183

Bradstock, J. E. Williams, and A. M. Grill, editors. Flammable Australia: the fire regimes and biodiversity of a continent. Cambridge University Press, Cambridge, UK.

Russell-Smith, J., P. J. Stanton, A. C. Edwards, and P. J. Whitehead. 2004a. Rain forest invasion of eucalypt-dominated woodland savanna, Iron Range, northeastern Australia: II. Rates of landscape change. Journal of Biogeography 31:1305-1316.

Russell-Smith, J., P. J. Stanton, P. J. Whitehead, and A. Edwards. 2004b. Rain forest invasion of eucalypt-dominated woodland savanna, Iron Range, north-eastern Australia: I. Successional processes. Journal of Biogeography 31:1293-1303.

Saulei, S., and M. Swaine. 1988. Rain forest seed dynamics during duccession at Gogol, Papua New Guinea. The Journal of Ecology 76:1133-1152.

Saunders, D. A., R. Hobbs, and C. Margules. 1991. Biological consequences of ecosystem fragmentation: A review. Conservation Biology 5:18-32.

Scariot, A. 2000. Seedling mortality by litterfall in Amazonian forest fragments. Biotropica 32:662-669.

Schedlbauer, J. L., B. Finegan, and K. L. Kavanagh. 2007. Rain forest structure at forest-pasture edges in northeastern Costa Rica. Biotropica 39:578-584.

Schneider, C., and C. Moritz. 1999. Rainforest refugia and evolution in Australia's wet tropics. Proceedings: Royal Society of London, Biological Sciences 266:191- 196.

Schnitzer, S., J. Dalling, and W. Carson. 2000. The impact of lianas on tree regeneration in tropical forest canopy gaps: evidence for an alternative pathway of gap-phase regeneration. Journal of Ecology 88:655-666.

Schnitzer, S., M. Kuzee, and F. Bongers. 2005. Disentangling above-and below- ground competition between lianas and trees in a tropical forest. Journal of Ecology 93:1115-1125.

Schnitzer, S. A. 2005. A mechanistic explanation for global patterns of liana abundance and distribution. American Naturalist 166:262-276.

Schnitzer, S. A., and F. Bongers. 2002. The ecology of lianas and their role in forests. Trends in Ecology & Evolution 17:223-230.

Schnitzer, S. A., and W. P. Carson. 2001. Treefall gaps and the maintenance of species diversity in a tropical forest. Ecology 82:913-919.

Schnitzer, S. A., M. P. E. Parren, and F. Bongers. 2004. Recruitment of lianas into logging gaps and the effects of pre-harvest climber cutting in a lowland forest in Cameroon. Forest Ecology and Management 190:87-98.

184

Schowalter, T. D., Y. Zhang, and R. A. Progar. 2005. Canopy arthropod response to density and distribution of green trees retained after partial harvest. Ecological Applications 15:1594-1603.

Schupp, E., H. Howe, C. Augspurger, and D. Levey. 1989. Arrival and survival in tropical treefall gaps. Ecology 70:562-564.

Scougall, S. A., J. D. Majer, and R. J. Hobbs. 1993. Edge effects in grazed and ungrazed Western Australian wheatbelt remnants in relation to ecosystem reconstruction. Nature Conservation 3:163–178.

Shugart, H., M. Hopkins, I. P. Burgess, and A. Mortlock. 1980. The development of a succession model for subtropical rain forest and its application to assess the effects of timeber harvest at wiangaree state forest, New South Wales. Journal of Environmental Management 11:243-265.

Singh, G., A. P. Kershaw, and R. Clark. 1981. Quaternary vegetation and fire history in Australia. Pages 23-54 in A. M. Gill, R. H. Groves, and I. R. Noble, editors. Fire and the Australian Biota. Australian Academy of Science, Canberra

Singhakumara, B., R. Uduporuwa, and P. Ashton. 2000. Soil seed banks in relation to light and topographic position of a hill dipterocarp forest in Sri Lanka. Biotropica 32:190-196.

Sizer, N., and E. Tanner. 1999. Responses of woody plant seedlings to edge formation in a lowland tropical rainforest, Amazonia. Biological Conservation 91:135- 142.

Smart, S. M., R. G. H. Bunce, R. Marrs, M. Leduc, L. G. Firbank, L. C. Maskell, W. A. Scott, K. Thompson, and K. J. Walker. 2005. Large-scale changes in the abundance of common higher plant species across Britain between 1978, 1990 and 1998 as a consequence of human activity: Tests of hypothesised changes in trait representation. Biological Conservation 124:355-371.

Smith, J.M.B, and I. Guyer. 1983. Rainforest-eucalypt forest interactions and the relevance of the biological nomad concept. Australian Journal of Ecology 8:55-60.

Smith, R., J. Nichols, and J. Vanclay. 2005. Dynamics of tree diversity in undisturbed and logged subtropical rainforest in Australia. Biodiversity and Conservation 14:2447-2463.

Sodhi, N., L. Koh, B. Brook, and P. Ng. 2004. Southeast Asian biodiversity: an impending disaster. Trends in Ecology & Evolution 19:654-660.

Spellerberg, I. F. 1998. Ecological effects of roads and traffic: a literature review. Global Ecology and Biogeography Letters 7:317-333.

Spss Inc. 2004. SPSS 13.0 for Windows. SPSS Inc., Chicago IL.

185

Stevens, G. 1987. Lianas as structural parasites: The bursera simaruba example. Ecology 68:77-81.

Stevens, N. C. 1977. Geology and landforms. Pages 1-6. The Border Ranges : A Land Use Conflict in Regional Perspective. the Royal Society of Queensland, Queensland.

Stewart, B. C. 1995. Seed and Seedling Ecology in a Secondary Subtropical Rainforest. Page 245. Southern Cross University. Southern Cross University.

Swaine, M. D., and T. C. Whitmore. 1988. On the definition of ecological species groups in tropical rain forests. Plant Ecology 75:81-86.

Symington, C. F. 1933. The study of secondary growth on rain forest sites in Malaya. Malayan Forester 2:107-117.

Tabanez, A. A. J., V. M. Viana, and D. Da S. 1997. Fragmentation and edge effect consequences on structure, diversity and sustainability of a plateau forest fragment in Piracicaba SP. Revista Brasileira de Biologia 57:47-60.

Tang, Y., S. L. Boulter, and R. L. Kitching. 2003. Heat and smoke effects on the germination of seeds from soil seed banks across forest edges between subtropical rainforest and eucalypt forest at Lamington National Park. Australian Journal of Botany 51:227-237.

Tang, Y., M. Cao, and X. Fu. 2006. Soil Seedbank in a Dipterocarp Rain Forest in Xishuangbanna, Southwest China. Biotropica 38:328-333.

Terborgh, J., L. Lopez, P. Nunez, M. Rao, G. Shahabuddin, G. Orihuela, M. Riveros, R. Ascanio, G. H. Adler, and T. D. Lambert. 2001. Ecological meltdown in predator-free forest fragments. Science 294:1923-1926.

Terborgh, J., and G. Nunez-Iturri. 2006. Disperser-free tropical forests await an unhappy fate. Pages 241–252 in W. F. Laurance, and C. A. Peres, editors. Emerging Threats to Topical Forests. University of Chicago press, Chicago.

Tilman, D., J. Fargione, B. Wolff, C. D'antonio, A. Dobson, R. Howarth, D. Schindler, W. H. Schlesinger, D. Simberloff, and D. Swackhamer. 2001. Forecasting agriculturally driven global environmental change. Science 292:281-284.

Tilman, D., R. M. May, C. L. Lehman, and M. A. Nowak. 1994. Habitat destruction and the extinction debt. Nature 371:65-66.

Toh, I., M. Gillespie, and D. Lamb. 1999. The role of isolated trees in facilitating tree seedling recruitment at a degraded sub-tropical rainforest site. Restoration Ecology 7:288-297.

Turner, I. 1996. Species loss in fragments of tropical rain forest: A review of the evidence. The Journal of Applied Ecology 33:200-209.

186

Turner, I., and R. Corlett. 1996. The conservation value of small, isolated fragments of lowland tropical rain forest. Trends in Ecology and Evolution 11:330-333.

Turton, S., and G. Duff. 1992. Light environments and floristic composition across an open forest-rainforest boundary in northeastern Queensland. Australian Journal of Ecology 17:415-523.

Turton, S., and G. Sexton. 1996. Environmental gradients across four rainforest-open forest boundaries in northeastern Queensland. Australian Journal of Ecology 21:245-254.

Turton, S. M., and H. J. Freiburger. 1997. Edge and aspect effects on the microclimate of a small tropical forest remnant on the Atherton Tableland, northeastern Australia. Pages 45–54 in W. F. Laurance, and R. O. Bierregaard, editors. Tropical Forest Remnants: Ecology, Management and Conservation of Fragmented Communities. University of Chicago Press, Chicago.

Uhl, C. 1987. Factors controlling succession following slash-and-brn ariculture in Amazonia. The Journal of Ecology 75:377-407.

Uhl, C., K. Clark, H. Clark, and P. Murphy. 1981. Early plant succession after cutting and burning in the Upper Rio Negro region of the Amazon Basin. The Journal of Ecology 69:631-649.

Uhl, C., K. Clark, N. Dezzeo, and P. Maquirino. 1988. Vegetation dynamics in Amazonian treefall gaps. Ecology 69:751-763.

Unwin, G. 1989. Structure and composition of the abrupt rainforest boundary in the Herberton Highland, north Queensland. Australian Journal of Botany 37:413– 428.

Vandermeer, J., I. G. Cerda, D. Boucher, I. Perfecto, and J. Ruiz. 2000. Hurricane disturbance and tropical tree species diversity. Science 290:788-791.

Vazquez-Yanes, C., and A. Orozco-Segovia. 1993. Patterns of seed longevity and germination in the tropical rainforest. Annual Review of Ecology and Systematics 24:69-87.

Vellend, M., K. Verheyen, H. Jacquemyn, A. Kolb, H. Van Calster, G. Peterken, and M. Hermy. 2006. Extinction debt of forest plants persists for more than a century following habitat fragmentation. Ecology 87:542-548.

Vidal, E., J. Johns, J. Gerwing, P. Barreto, and C. Uhl. 1997. Vine management for reduced-impact logging in eastern Amazonia. Forest Ecology and Management 98:105-114.

Wahungu, G., C. Catterall, and M. Olsen. 2002. Seedling predation and growth at a rainforest-pasture ecotone, and the value of shoots as seedling analogues. Forest Ecology and Management 162:251-260.

187

Wahungu, G. M., C. P. Catterall, and M. F. Olsen. 1999. Selective herbivory by red- necked pademelon Thylogale thetis at rainforest margins: factors affecting predation rates. Australian Journal of Ecology 24:577-586.

Waller, D. M. 1993. How does mast-fruiting get started. Trends in Ecology and Evolution 8:122-123.

Walsh, K. J. E., and B. F. Ryan. 2000. Tropical cyclone intensity increase near Australia as a result of climate change. Journal of Climate 13:3029-3036.

Wang, B., and T. Smith. 2002. Closing the seed dispersal loop. Trends in Ecology & Evolution 17:379-386.

Wang, J. 1996. Seed Dynamics Across Rainforest-Eucalypt Forest Boundary of Southeastern Queensland Thesis (Ph.D., page 137), University of Queensland, Brisbane.

Wang, L. 1997. The soil seed bank and understorey regeneration in Eucalyptus regnans forest, Victoria. Australian Journal of Ecology 22:404-411.

Weathers, K. C., M. L. Cadenasso, and S. T. A. Pickett. 2001. Forest edges as nutrient and pollutant concentrators: potential synergisms between fragmentation, forest canopies, and the atmosphere. Conservation Biology 15:1506-1514.

Weaver, D. B. 2002. Hard-core ecotourists in Lamington National Park, Australia. Journal of Ecotourism 1:19-35.

Webb, C. O., and D. R. Peart. 2000. Habitat associations of trees and seedlings in a Bornean rain forest. Journal of Ecology 88:464-478.

Webb, L. 1958. Cyclones as an ecological factor in tropical lowland rainforest, North Queensland. Australian Journal of Botany 6:220-228.

Webb, L. 1968. Environmental relationships of the structural types of Australian rain forest vegetation. Ecology 49:296-311.

Webb, L., J. Tracey, and W. Williams. 1972. Regeneration and pattern in the subtropical rain forest. The Journal of Ecology 60:675-695.

Webb, L. J. 1978. A general classification of Australian rainforests. Australian Plants 9:349-363.

Webb, L. J., and J. G. Tracey. 1981. Australian rainforests: patterns and change. Pages 605-694 in A. Keast, editor. Ecological Biogeography of Australia. Dr. W. Junk, The Hague, Netherlands.

Webb, L. J., J. G. Tracy, and W. T. Williams. 1984. A floristic framework of Australian rainforests. Australian Journal of Ecology 9:169-198.

White, E., N. Tucker, N. Meyers, and J. Wilson. 2004. Seed dispersal to revegetated

188

isolated rainforest patches in North Queensland. Forest Ecology and Management 192:409-426.

Whitmore, T. C. 1983. Secondary succession from seed in tropical rain forests. Forestry Abstracts 44:767-779.

Whitmore, T. C. 1989. Canopy gaps and the two major groups of forest trees. Ecology 70:536-538.

Whitmore, T. C. 1996. A review of some aspects of tropical rain forest seedling ecology with suggestions for further enquiry. Pages 3–30 in T. Whitmore, and M. Swaine, editors. The Ecology of Tropical Forest Tree Seedlings. UNESCO and Parthenon Publishing group, Paris.

Whitmore, T. C. 1997. Tropical forest disturbance, disappearance, and species loss. Pages 3–12 in W. F. Laurance, and R. O. Bierregaard, Jr., editors. Tropical forest remnants: ecology, management, and conservation of fragmented communities. University of Chicago Press, Chicago.

Whitmore, T. C. 1998a. An Introduction to Tropical Rain Forest. Oxford University Press, New York.

Whitmore, T. C. 1998b. Potential impact of climatic change on tropical rain forest seedlings and forest regeneration. Climatic Change 39:429-438.

Whitmore, T. C., and J. A. Sayer. 1992. Deforestation and species extinction in tropical moist forests. Pages 1–14 in T. C. Whitmore, and J. A. Sayer, editors. Tropical Deforestation and Species Extinction. Chapman and Hall, London.

Wijdeven, S., and M. Kuzee. 2000. Seed availability as a limiting factor in forest recovery processes in Costa Rica. Restoration Ecology 8:414-424.

Williams-Linera, G. 1990a. Origin and early development of forest edge vegetation in Panama. Biotropica 22:235-241.

Williams-Linera, G. 1990b. Vegetation structure and environmental conditions of forest edges in Panama. The Journal of Ecology 78:356-373.

Williams-Linera, G. 2002. Tree species richness complementarity, disturbance and fragmentation in a Mexican tropical montane cloud forest. Biodiversity and Conservation 11:1825-1843.

Williams-Linera, G., V. Dominguez-Gastelu, and M. Garcia-Zurita. 1998. Microenvironment and floristics of different edges in a fragmented tropical rainforest. Conservation Biology 12:1091-1102.

Williams, A. A. J., D. J. Karoly, and N. Tapper. 2001. The sensitivity of Australian fire danger to climate change. Climatic Change 49:171-191.

Williams, G., T. Evans, and A. Museum 1993. Hidden Rainforests: Subtropical

189

Rainforests and Their Invertebrate Biodiversity. New South Wales University Press, Australia.

Williams, J., and G. Harden 2000. Rainforest Climbing Plants: A Field Guide to The Rainforest Climbing Plants of New South Wales Using Vegetative Characters. Botany Department, University of New England, Armidale, New Souths Wales.

Williams, S. E. 2003. Climate change in Australian tropical rainforests: an impending environmental catastrophe. Proceedings: Biological Sciences 270:1887-1892.

Williams, S. E., and D. W. Hilbert. 2006. Climate change threats to the biodiversity of tropical rainforests in Australia in W. Laurance, and C. A. Peres, editors. Emerging Threats to Tropical Forests. Chicago University Press, Chicago, IL, USA.

Williams, S. E., and H. Marsh. 1998. Changes in small mammal assemblage structure across a rain forest/open forest ecotone. Journal of Tropical Ecology 14:187- 198.

Williamson, G. B., and K. Ickes. 2002. Mast fruiting and ENSO cycles- does the cue betray a cause? Oikos 97:459-461.

Willis, K. J., L. Gillson, and T. M. Brncic. 2004. How “virgin” is virgin rainforest. Science 304:402-403.

Willson, M., and F. Crome. 1989. Patterns of seed rain at the edge of a tropical Queensland rain forest. Journal of Tropical Ecology 5:301-308.

Wright, S.J. 2002. Plant diversity in tropical forests: a review of mechanisms of species coexistence. Oecologia 130:1-14.

Wright, S.J., and H. Muller-Landau. 2006a. The future of tropical forest species. Biotropica 38:287.

Wright, S. J. 2005. Tropical forests in a changing environment. Trends in Ecology & Evolution 20:553-560.

Wright, S. J., O. Calderon, A. Hernandez, and S. Paton. 2004. Are lianas increasing in importance in tropical forests? A 17-year record from Panama. Ecology 85:484-489.

Wright, S. J., and H. C. Muller-Landau. 2006b. The uncertain future of tropical forest species. Biotropica 38:443-445.

Wright, S. J., H. Zeballos, I. Dominguez, M. M. Gallardo, M. C. Moreno, and R. Ibanez. 2000. Poachers alter mammal abundance, seed dispersal, and seed predation in a Neotropical forest. Conservation Biology 14:227-239.

Wunderle, J. 1997. The role of animal seed dispersal in accelerating native forest regeneration on degraded tropical lands. Forest Ecology and Management

190

99:223-235.

Young, K. R., J. J. Ewel, and B. J. Brown. 1987. Seed dynamics during forest succession in Costa Rica. Plant Ecology 71:157-173.

191

192

APPENDIX 1

List of plant species and their occurrence in different forms in the present study

193

194

Appendix 1 List of plant species and their occurrence in different forms in the present study (tree, liana, seedling and seeds in soil seed bank). Nomenclature and plant life forms (G: grass, F: forb, V: Vine, LS: low shrub, TS: tall shrub, LT: low tree, TT: tall tree) followed McDonald and Thomas (1990) and recent changes in species names were checked using the online flora of New South Wale

(http:/H / plantnet.rbgsyd.nsw.gov.au).H Forms recorded in present study Species Life form Family Tree Liana seedling seed bank Acanthaceae Pseuderanthemum variabile (R.Br.) Radlk. F * Agavaceae Cordyline rubra Otto & A.Dietr. LS * Akaniaceae Akania bidwillii (Hogg) Mabb. LT * Alangiaceae Alangium villosum (Blume) Wangerin subsp. polyosmoides (F.Muell.) Bloemb. LT * * * Amaranthaceae Amaranthus viridis L. F * Deeringia amaranthoides (Lam.) Merr. F * Nyssanthes diffusa R.Br. F * Euroschinus falcatus Hook.f. var. falcatus LT/ TT * Rhodosphaera rhodanthema (F.Muell.) Engl. LT/ TT * Annonaceae Meiogyne stenopetala (F.Muell.) Heusden subsp. stenopetala TS * Melodorum leichhardtii (F.Muell.) Benth. V * Apiaceae Daucus glochidiatus (Labill.) Fisch., C.A.Mey. & Ave-Lall. F * Hydrocotyle bonariensis Lam. F * Hydrocotyle pedicellosa F.Muell. F * Hydrocotyle peduncularis R.Br. ex A.Rich. F * Apocynaceae Alyxia ruscifolia R.Br. TS * Carissa ovata R.Br. LS * Melodinus australis (F.Muell.) Pierre V * * Parsonsia longipetiolata J.B.Williams V * Parsonsia straminea (R.Br.) F.Muell. V * Parsonsia velutina R.Br. V * * Araliaceae Cephalaralia cephalobotrys (F.Muell.) Harms V * * Polyscias elegans (C.Moore & F.Muell.) Harms LT/ TT * * * Polyscias murrayi (F.Muell.) Harms LT/ TT * * Araucariaceae Araucaria cunninghamii Aiton ex A.Cunn. var. cunninghamii TT * * Arecaceae Archontophoenix cunninghamiana (H.Wendl.) H.Wendl. & Drude LT/ TT * 195 Linospadix monostachya (Mart.) H.Wendl. LS/ TS *

196 Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Asclepiadaceae Gomphocarpus physocarpus E.Mey. LS * Marsdenia rostrata R.Br. V * Asteraceae Ageratina adenophora (Spreng.) R.M.King & H.Rob. F/ LS * Ageratina riparia (Regel) R.M.King & H.Rob. F/ LS * Bidens pilosa L. F * Cirsium vulgare (Savi) Ten. F * Conyza canadensis (L.) Cronquist F * Asteraceae Crassocephalum crepidioides (Benth.) S.Moore F * Emilia sonchifolia (L.) DC. F * Euchiton involucratus (G.Forst.) Holub F * Galinsoga parviflora Cav. F * Gamochaeta spicata (Lam.) Cabrera F * Hypochaeris radicata L. F * Olearia heterocarpa S.T.Blake TS/ LS * Senecio bipinnatisectus Belcher F * Senecio linearifolius A.Rich. F * Senecio madagascariensis Poir. F * Sigesbeckia orientalis L. F * Sonchus asper (L.) Hill F * Youngia japonica (L.) DC. F * * * Bignoniaceae Pandorea baileyana (Maiden & R.T.Baker) Steenis V * Pandorea jasminoides (Lindl.) K.Schum. V * Pandorea pandorana (Andrews) Steenis V * * Boraginaceae Ehretia acuminata R.Br. var. acuminata TT * Brassicaceae Lepidium virginicum L. F * Caesalpiniaceae Caesalpinia subtropica Pedley V * Campanulaceae Lobelia trigonocaulis F.Muell. F * Capparaceae Capparis arborea (F.Muell.) Maiden LT * * Caryophyllaceae Cerastium glomeratum Thuill. F * Stellaria flaccida Hook. F * * Celastraceae Celastrus australis Harv. & F.Muell. ex F.Muell. V/ LS *

Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Celastraceae Celastrus subspicata Hook. V/ LS * * Denhamia celastroides (F.Muell.) Jessup LT * * Elaeodendron australe Vent. var. australe LT * * Maytenus disperma (F.Muell.) Loes. LT * Chenopodiaceae Chenopodium carinatum R.Br. F * Einadia hastata (R.Br.) A.J.Scott F * Commeliaceae Commelina cyanea R.Br. F * Cucurbitaceae Sicyos australis Endl. V * Cunoniaceae Caldcluvia paniculosa (F.Muell.) Hoogland TT * * * Geissois benthamii F.Muell. TT * Pseudoweinmannia lachnocarpa (F.Muell.) Engl. TT * * * Cyatheaceae Cyathea leichhardtiana (F.Muell.) Copel. TS/ LT * Cyperaceae Bulbostylis barbata (Rottb.) C.B.Clarke G * Carex inversa R.Br. G * Carex pumila Thunb. ex Murray G * Cyperus brevifolius (Rottb.) Hassk. G * Cyperus gracilis R.Br. G * Cyperus rotundus L. G * Cyperus sesquiflorus (Torr.) Mattf. & Kuek. G * Cyperus tetraphyllus R.Br. G * Fimbristylis sp G * Dilleniaceae Hibbertia scandens (Willd.) Gilg var. glabra Maiden V * * Dioscoreaceae Dioscorea transversa R.Br. V/ F * Ebenaceae Diospyros australis (R.Br.) Hiern TS/ LT * * Diospyros pentamera (Woolls & F.Muell.) F.Muell. TT * * Elaeocarpaceae Elaeocarpus obovatus G.Don TT * * * Elaeocarpus reticulatus Sm. TS * Sloanea woollsii F.Muell. TT * Euphorbiaceae Actephila lindleyi (Steud.) Airy Shaw TS * *

197 Alchornea ilicifolia (J.Sm.) Muell.Arg. TS * * Baloghia inophylla (G.Forst.) P.S.Green LT * *

198 Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Euphorbiaceae Breynia oblongifolia (Muell.Arg.) Muell.Arg. var. oblongifolia LS/ TS * Claoxylon australe Baill. ex Muell.Arg. TS * Cleistanthus cunninghamii (Muell.Arg.) Muell.Arg. TS * * Croton verreauxii Baill. TS * * Drypetes deplanchei (Brongn. & Gris) Merr. LT/ TT * * Excoecaria dallachyana (Baill.) Benth. LT/ TT * Homalanthus nutans (G.Forst.) Guill LT * * Mallotus philippensis (Lam.) Muell.Arg. LT * * Petalostigma pubescens Domin G * Eupomatia bennettii F.Muell. LS * Eupomatia laurina R.Br. TS/ LT * * Fabaceae Austrosteenisia blackii (F.Muell.) R.Geesink V * Austrosteenisia glabristyla Jessup V * Derris involuta (Sprague) Sprague V * * Kennedia rubicunda (Schneev.) Vent. V * Fabaceae Trifolium campestre Schreb. var. campestre F * * Flacourtiaceae Casearia multinervosa C.T.White & Sleumer ex Sleumer TS/ LT * * Scolopia braunii (Klotzsch) Sleumer LT.TT * * Streptothamnus moorei F.Muell. V * * Geraniaceae Geranium solanderi Carolin var. solanderi F * Grossulariaceae Polyosma cunninghamii Benn. LT/ TT * * Quintinia verdonii F.Muell. LT * * * Icacinaceae Citronella moorei (F.Muell. ex Benth.) R.A.Howard TT * * Pennantia cunninghamii Miers TT * Iridaceae Sisyrinchium iridifolium Kunth G * Juncaceae Juncus flavidus L.A.S.Johnson G * Lamiaceae Clerodendrum floribundum R.Br. TS/ LT * * Plectranthus argentatus S.T.Blake F/ LS * Vitex lignum-vitae A.Cunn. ex Schauer TT * * * Lauraceae Beilschmiedia elliptica C.T.White & W.D.Francis LT/ TT * Cinnamomum oliveri F.M.Bailey TT * *

Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Lauraceae Cinnamomum virens R.T.Baker TT * * Cryptocarya erythroxylon Maiden & Betche ex Maiden TT * * Cryptocarya foveolata C.T.White & W.D.Francis TT * Cryptocarya bidwillii Meisn. LT * Cryptocarya obovata R.Br. TT * * Cryptocarya triplinervis R.Br. LT * * Endiandra muelleri Meisn. subsp. muelleri TT * * Endiandra pubens Meisn. LT * Litsea reticulata (Meisn.) F.Muell. TT * * Neolitsea australiensis Kosterm. LT/ TT * * Neolitsea dealbata (R.Br.) Merr. TS/ LT * * Meliaceae Anthocarapa nitidula (Benth.) T.D.Penn. ex Mabb. LT/ TT * * Dysoxylum fraserianum (A.Juss.) Benth. TT * * Dysoxylum mollissimum Blume subsp. molle (Miq.) Mabb. TT * * Dysoxylum rufum (A.Rich.) Benth. LT/ TT * * Owenia cepiodora F.Muell. TT * Synoum glandulosum (Sm.) A.Juss. var. glandulosum LT/ TT * Menispermaceae Legnephora moorei (F.Muell.) Miers V * * Sarcopetalum harveyanum F.Muell. V * Menispermaceae Stephania japonica (Thunb.) Miers var. discolor (Blume) Forman V * * Mimosaceae Acacia melanoxylon R.Br. LT/ TT * * Archidendron grandiflorum (Sol. ex Benth.) I.C.Nielsen LT/ TT * * Pararchidendron pruinosum (Benth.) I.C.Nielsen var. pruinosum LT * * Monimiaceae Daphnandra tenuipes Perkins LT/ TT * * Doryphora sassafras Endl. TT * * Palmeria scandens F.Muell. V * Wilkiea austroqueenslandica Domin TS * * Wilkiea huegeliana (Tul.) A.DC. TS/ LT * * Wilkiea macrophylla (A.Cunn.) A.DC. TS *

199 Moraceae Ficus coronata Spin LT * Ficus superba (Miq.) Miq. var. henneana (Miq.) Corner TT * *

200 Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Moraceae Ficus watkinsiana F.M.Bailey TT * * * Maclura cochinchinensis (Lour.) Corner V/ TS * * Streblus brunonianus (Endl.) F.Muell. LT/ TT * Trophis scandens (Lour.) Hook. & Arn. subsp. scandens V * Myrsinaceae Embelia australiana (F.Muell.) F.M.Bailey V/ LS * * Myrtaceae Acmena ingens (F.Muell. ex C.Moore) Guymer & B.Hyland TT * * Acmena smithii (Poir.) Merr. & L.M.Perry LT/ TT * * Angophora subvelutina F.Muell. LT * Archirhodomyrtus beckleri (F.Muell.) A.J.Scott TS/ LT * * Decaspermum humile (G.Don) A.J.Scott LT * Eucalyptus acmenoides Schauer TT * Eucalyptus biturbinata L.A.S.Johnson & K.D.Hill TT * Eucalyptus campanulata R.T.Baker & H.G.Sm. TT * Eucalyptus microcorys F.Muell. TT * Eucalyptus notabilis Maiden TS * Eucalyptus quadrangulata H.Deane & Maiden TT * Gossia acmenoides (F.Muell.) N.Snow & Guymer LT * * Gossia bidwillii (Benth.) N.Snow & Guymer LT * Lenwebbia prominens N.Snow & Guymer LT * * * Lophostemon confertus (R.Br.) Peter G.Wilson & J.T.Waterh. TT * * Pilidiostigma glabrum Burret TS/ LT * Rhodamnia argentea Benth. TT * * Rhodamnia rubescens (Benth.) Miq. LT/ TT * Myrtaceae Rhodomyrtus psidioides (G.Don) Benth. LT * * * Syzygium australe (H.L.Wendl. ex Link) B.Hyland LT/ TT * * Syzygium crebrinerve (C.T.White) L.A.S.Johnson TT * * * Mysiaceae Myrsine subsessilis F.Muell. subsp. subsessilis LS/ TS * * Oleaceae Jasminum dallachii F.Muell. V * Notelaea johnsonii P.S.Green TS/ LT * * Notelaea longifolia forma glabra P.S.Green TS/ LT * Olea paniculata R.Br. TT *

Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Oxalidaceae Oxalis corniculata L. F * Passifloraceae Passiflora herbertiana Ker Gawl. V * Phormiaceae Dianella caerulea Sims var. caerulea F * Phytolaccaceae Phytolacca octandra L. F/ LS * Piperaceae Piper novae-hollandiae Miq. V * Pittosporaceae Pittosporum multiflorum (A.Cunn. ex Loudon) L.W.Cayzer et al. LT * Hymenosporum flavum (Hook.) F.Muell. LT * Pittosporum lancifolium (F.M.Bailey) L.W.Cayzer, Crisp & I.Telford TS/ LT * * Pittosporum revolutum W.T.Aiton LS * Pittosporum rhombifolium A.Cunn. ex Hook. LT/ TT * Plantaginaceae Plantago debilis R.Br. F * Poaceae Axonopus compressus (Sw.) P.Beauv. G * Briza minor L. G * Bromus diandrus Roth G * Dichelachne inaequiglumis (Hack. ex Cheeseman) Edgar & Connor G * Digitaria sanguinalis (L.) Scop. G * Digitaria violascens Link G * Elymus scaber (R.Br.) A.Le G * Eragrostis sp G * Microlaena stipoides (Labill.) R.Br. var. stipoides G * Oplismenus aemulus (R.Br.) Roem. & Schult. G * Panicum decompositum R.Br. G * Paspalum urvillei Steud. G * Pennisetum clandestinum Hochst. ex Chiov. G * Poa labillardieri Steud. var. labillardieri G * Setaria pumila (Poir.) Roem. & Schult. G * Tragus australianus S.T.Blake G * * Polygonaceae Muehlenbeckia gracillima Meisn. F/ V * Proteaceae Helicia glabriflora F.Muell. LT * *

201 Orites excelsus R.Br. LT/ TT * * Stenocarpus salignus R.Br. LT/ TT * *

202 Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Proteaceae Stenocarpus sinuatus (Loudon) Endl. LT/ TT * Triunia youngiana (C.Moore & F.Muell. ex F.Muell.) L.A.S.Johnson & B.G.Briggs LS/ TS * Ranunculaceae Clematis glycinoides DC. V * * Rhamnaceae Alphitonia excelsa (A.Cunn. ex Fenzl) Reissek ex Benth. LT/ TT * * * Emmenosperma alphitonioides F.Muell. TT * Rosaceae Duchesnea indica (Jacks.) Th.Wolf F * Rubus moluccanus L. var. trilobus A.R.Bean LS/ V * Rubus moorei F.Muell. V * * Rubus nebulosus A.R.Bean V * Rubus parvifolius L. F/ V * Rubus rosifolius Sm. var. rosifolius LS/ V * * Rubiaceae Atractocarpus benthamianus (F.Muell.) Puttock subsp. benthamianus TS/ LT * * * Atractocarpus chartaceus (F.Muell.) Puttock TS * * Psydrax odorata (G.Forst.) A.C.Sm. & S.P.Darwin TS/ LT * Galium migrans Ehrend. & McGill. F * Hodgkinsonia ovatiflora F.Muell. LT * Morinda jasminoides A.Cunn. ex Hook. V * * Psychotria daphnoides A.Cunn. var. pubescens F.M.Bailey LS * Psychotria loniceroides Sieber ex DC. LS * Psychotria simmondsiana F.M.Bailey var. exigua F.M.Bailey LS * * * Rutaceae Acronychia laevis J.R.Forst. & G.Forst. LT * * * Acronychia oblongifolia (A.Cunn. ex Hook.) Endl. ex Heynh. LT * * * Acronychia pauciflora C.T.White TS/ LT * * * Acronychia pubescens (F.M.Bailey) C.T.White LT * * * Acronychia suberosa C.T.White LT/ TT * * * Flindersia australis R.Br. TT * * Geijera salicifolia Schott LT/ TT * * * Halfordia kendack (Montrouz.) Guillaumin LT/ TT * * Melicope micrococca (F.Muell.) T.G.Hartley LT/ TT * * * Acronychia octandra (F.Muell.) T.G.Hartley TT * * *

Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Rutaceae Citrus australasica F.Muell. TS/ LT * * Pentaceras australe (F.Muell.) Benth. LT/ TT * * Leionema elatius (F. Muell.) Paul G.Wilson subsp. elatius TS * Sarcomelicope simplicifolia (Endl.) T.G.Hartley subsp. simplicifolia LT * * Zanthoxylum brachyacanthum F.Muell. LT Sapindaceae Alectryon subcinereus (A.Gray) Radlk. LT/ TT * * Arytera distylis (F.Muell. ex Benth.) Radlk. TS/ LT * * Arytera divaricata F.Muell. TS/ LT * * Atalaya multiflora Benth. LT * * Cupaniopsis flagelliformis (F.M.Bailey) Radlk. var. australis S.T.Reynolds LT * * Cupaniopsis newmanii S.T.Reynolds TS * * Cupaniopsis serrata (F.Muell.) Radlk. LT * * Diploglottis australis (G.Don) Radlk. LT/ TT * * Elattostachys nervosa (F.Muell.) Radlk. LT/ TT * * Elattostachys xylocarpa (A.Cunn. ex F.Muell.) Radlk. LT/ TT * * Guioa semiglauca (F.Muell.) Radlk. LT/ TT * * Harpullia alata F.Muell. TS * Harpullia hillii F.Muell. LT * * Jagera pseudorhus (A.Rich.) Radlk. forma pseudorhus LT/ TT * * Lepiderema pulchella Radlk. LT * * Mischocarpus anodontus (F.Muell.) Radlk. TS * * Rhysotoechia bifoliolata Radlk. subsp. bifoliolata LT/ TT * * Sarcopteryx stipata (F.Muell.) Radlk. LT/ TT * * Toechima tenax (A.Cunn. ex Benth.) Radlk. LT * Sapotaceae Pouteria australis (R.Br.) Baehni TT * * Pouteria myrsinoides (A.Cunn. ex Benth.) Baehni LT * * Pouteria pohlmaniana (F.Muell.) Baehni LT/ TT * * Scrophulariaceae Euphrasia collina R.Br. F * Mazus pumilio R.Br. F *

203 Veronica plebeia R.Br. F * Eustrephus latifolius R.Br. ex Ker Gawl. V *

204 Forms recorded in present study Family Species Life form Tree Liana seedling seed bank Solanaceae Duboisia myoporoides R.Br. LT * Physalis peruviana L. F * Solanum aviculare G.Forst. TS * Solanum mauritianum Scop. TS * Solanum americanum Mill. F/ LS * Solanum nigrum L. F * Solanum semiarmatum F.Muell. LS * * Solanum stelligerum Sm. LS * Sterculiaceae Argyrodendron actinophyllum (F.M.Bailey) Edlin subsp. actinophyllum TT * * Argyrodendron trifoliolatum F.Muell. TT * * Brachychiton acerifolius (A.Cunn. ex G.Don) Macarthur TT * * Brachychiton bidwillii LT * * Brachychiton discolor F.Muell. TT * * Surianaceae Guilfoylia monostylis (Benth.) F.Muell. LT * * Symplocaceae Symplocos thwaitesii F.Muell. LT * * Thymelaeaceae Phaleria chermsideana (F.M.Bailey) C.T.White LT * * Pimelea ligustrina Labill. subsp. ligustrina LS/ TS * Ulmaceae Aphananthe philippinensis Planch. LT/ TT * * * Trema tomentosa (Roxb.) Hara var. viridis (Planch.) Hewson LS/ TS * Urticaceae Dendrocnide excelsa (Wedd.) Chew TT * * * Urtica incisa Poir. F * Verbenaceae Lantana camara L. var. camara TS * Verbena rigida Spreng. F * Violaceae Viola betonicifolia Sm. subsp. betonicifolia F * * Vitaceae Cayratia eurynema B.L.Burtt V * * Cissus antarctica Vent. V * * Cissus hypoglauca A.Gray V * * Cissus sterculiifolia (F.Muell. ex Benth.) Planch. V * * Tetrastigma nitens (F.Muell.) Planch. V * * Winteraceae Tasmannia insipida R.Br. ex DC. LS *

APPENDIX 2

The distributions of lianas within liana carrying trees

205

206

Appendix 2 The distributions of lianas within liana carrying trees (trees with DBH>10 cm, the total tree and liana stems include those recorded at the reference rainforest plot and the three types of rainforest edges). Liana Species are: MEAU, Melodinus australis; AUGL, Austrosteenisia glabristyla; PAVE, Parsonsia velutina; TENI, Tetrastigma nitens; TRSC, Trophis scandens; CIAN, Cissus Antarctica; DEIN, Derris involuta; MOJA, Morinda jasminoides; PAPA, Pandorea pandorana; PAST, Parsonsia straminea; CESU, Celastrus subspicatus; CIST, Cissus sterculifolia; MELE, Melodorum leichhardtii. Stems Stems Lianas Other 24 Species recorded infested carried MEAU AUGL PAVE TENI TRSC CIAN DEIN MOJA PAPA PAST CESU CIST species Acronychia octandra 29 4 6 4 2 0 Acronychia suberosa 31 7 13 4 1 1 1 3 3 Alphitonia excelsa 16 7 21 5 1 2 2 11 Anthocarapa nitidula 25 8 20 7 2 3 2 1 2 3 Araucaria cunninghamii 23 5 12 1 2 1 4 2 1 1 Argyrodendron actinophyllum 82 24 48 10 12 3 4 3 2 3 1 2 8 Argyrodendron trifoliolatum 230 59 149 30 21 13 13 11 10 6 8 5 7 3 1 21 Atractocarpus benthamianus 71 17 28 15 3 2 3 1 1 3 Baloghia inophylla 158 32 55 14 17 7 1 2 4 2 1 7 Brachychiton acerifolius 24 4 6 1 1 1 1 2 Caldcluvia paniculosa 79 24 52 19 8 6 9 1 2 7 Daphnandra micrantha 15 1 4 3 1 Dendrocnide excelsa 32 7 14 4 4 2 1 1 1 1 Diospyros pentamera 94 19 35 11 4 3 8 1 1 1 2 2 2 Doryphora sassafras 52 8 16 6 4 2 1 1 2 Guilfoylia monostachyus 25 4 13 2 4 2 4 1 0 Guioa semiglauca 25 11 22 3 2 5 7 1 4 Halfordia kendack 22 9 21 11 3 2 1 1 3 0 Litsea reticulata 27 10 17 5 4 1 2 1 2 2 Melicope micrococca 32 7 12 1 4 4 2 1 Orites excelsus 51 9 16 7 5 2 2 0 Pentaceras australis 21 6 8 4 1 1 2 Phaleria chermsideana 18 1 1 1 0 Polyscias elegans 50 17 45 3 3 2 5 6 9 6 3 1 3 4 Pouteria australis 15 5 8 4 1 1 2 Pseudoweinmannia lachnocarpa 97 41 123 22 13 7 10 9 17 2 3 8 4 4 3 21 Sarcopteryx stipata 28 6 11 10 1 0 streblus brunonianus 36 3 4 3 1 Synoum glandulosum 18 5 9 6 1 1 0 207 Vitex lignum-vitae 34 17 42 4 3 7 4 3 5 1 5 1 1 8 Other 53 species 307 95 213 51 27 21 20 8 20 10 6 9 3 3 10 18

208

APPENDIX 3

Distributions of the total basal area of trees and lianas within the one hectare reference rainforest.

209

210

0 102030405060708090 0-50 50-100 100100-150 and above The distribution of total basal area of trees (m2/ha)

0 102030405060708090

0-0.5 0.5-1 1-1.5 1.5-2and above

The distribution of total basal area of lianas (m2/ha)

2 Appendix 3 Distributions of total basal area (mP /P ha) of trees (DBH ≥ 5 cm, above) and lianas (DBH ≥ 2cm, below) within the one hectare rainforest reference plot. A significant positive correlation was found between the total basal area of trees and lianas (Spearman Rank Correlation, P<0.05, Chapter 4), indicating the association of large lianas with large trees. The chart is generated using the chart function in Excel, based on the basal areas of trees and lianas in each of the 10×10 m squares.

211