Plant Ecology

, Volume 204, Issue 2, pp 231–246

Variation in exotic and native seed arrival and recruitment of bird dispersed species in subtropical forest restoration and regrowth

Authors

    • Alan Fletcher Research Station, Biosecurity Queensland, DPI and F and CRC for Australian Weed Management
  • Gabrielle Vivian-Smith
    • Alan Fletcher Research Station, Biosecurity Queensland, DPI and F and CRC for Australian Weed Management
  • Anna Barnes
    • Australian Rivers InstituteGriffith University
Article

DOI: 10.1007/s11258-009-9587-2

Cite this article as:
White, E., Vivian-Smith, G. & Barnes, A. Plant Ecol (2009) 204: 231. doi:10.1007/s11258-009-9587-2

Abstract

Invasive bird-dispersed plants often share the same suite of dispersers as co-occurring native species, resulting in a complex management issue. Integrated management strategies could incorporate manipulation of dispersal or establishment processes. To improve our understanding of these processes, we quantified seed rain, recruit and seed bank density, and species richness for bird-dispersed invasive and native species in three early successional subtropical habitats in eastern Australia: tree regrowth, shrub regrowth and native restoration plantings. We investigated the effects of environmental factors (leaf area index (LAI), distance to edge, herbaceous ground cover and distance to nearest neighbour) on seed rain, seed bank and recruit abundance. Propagule availability was not always a good predictor of recruitment. For instance, although native tree seed rain density was similar, and species richness was higher, in native plantings, compared with tree regrowth, recruit density and species richness were lower. Native plantings also received lower densities of invasive tree seed rain than did tree regrowth habitats, but supported a similar density of invasive tree recruits. Invasive shrub seed rain was recorded in highest densities in shrub regrowth sites, but recruit density was similar between habitats. We discuss the role of microsite characteristics in influencing post-dispersal processes and recruit composition, and suggest ways of manipulating these processes as part of an integrated management strategy for bird-dispersed weeds in natural areas.

Keywords

DispersalFleshy-fruited speciesInvasive plantRainforestSuccessionSeed bankSeed rainWeed management

Introduction

Bird-dispersed invasive species present a unique problem for land managers for a number of reasons. First, bird–plant interactions involve multiple species; most frugivores consume the fruits of many different plants (Jordano 1995; Wheelwright and Orians 1982), and the seeds of fleshy-fruited plants are usually dispersed by many bird species (Githiru et al. 2002; Stansbury and Vivian-Smith 2003). Second, birds can spread seed to isolated natural areas in which it can be difficult to locate and manage the invasion (Murphy et al. 2008; Westcott et al. 2008). Third, bird-dispersed weeds may provide benefits to native fauna through the provision of food and habitat (Date et al. 1991; Gosper and Vivian-Smith 2006), and to native flora by acting as pioneers and providing a suitable environment for seedling establishment (Neilan et al. 2006; Rodriguez 2006). Consequently, management of these species is often both complex and contentious.

The goal of management of invaded natural areas is usually to reduce weed abundance whilst facilitating establishment of native plant species (Hobbs and Norton 1996). To this end, as part of an integrated weed management and restoration program, it may be desirable to manage bird-dispersed weeds and native species simultaneously, through manipulation of species availability, site availability or species performance (sensu Pickett et al. 1987).

Gosper et al. (2005) suggest four approaches for manipulating species availability during the dispersal phase of invasive species’ lifecycle: (1) by reducing fruit production or quality, (2) through directing seed dispersal (for instance, by providing perch structures in areas unsuitable for establishment), (3) by prioritising removal of the major seed sources and (4) by facilitating establishment of native species to provide alternative resources for frugivores. Manipulation of such processes is complicated by the potential impacts of management strategies on native species; any action that affects dispersal of invasive species (e.g. perch sites construction) is likely to also impact native flora or fauna (Neilan et al. 2006; White et al. 2006). Furthermore, manipulating dispersal processes will only be successful if a species is dispersal-limited such as in the early stages of invasion.

Management strategies targeting establishment, in addition to dispersal, may be an effective option. Patterns of seed dispersal do not necessarily reflect patterns of seedling establishment, owing to variations in microsite suitability for germination and establishment (Jordano and Herrera 1995; McAlpine and Jesson 2008) and factors such as predation (Wenny 2000). Establishment can be manipulated by altering either site availability or species performance (sensu Pickett et al. 1987). In some situations, reduction of the availability of suitable establishment sites may be achieved by altering the disturbance regime (Geldenhuys 2004; Stone and Loope 1987). Species performance can be manipulated by simply removing or killing the invasive species (although this may not be possible in many situations), or through manipulation of biotic processes or interactions, for instance, by encouraging establishment of native species capable of out-competing the invader (Cummings et al. 2007).

Clearly, to reduce weed spread and establishment whilst promoting dominance by native bird-dispersed species, a thorough understanding of the dispersal and establishment processes of invasive and co-occurring native species is necessary.

The present study was conducted in subtropical eastern Australia where approximately two-thirds of native tree and shrub species have fruits dispersed by vertebrates, with birds being the most abundant and diverse vertebrate dispersers and where bird-dispersed weeds represent a quarter of the most invasive species (Gosper et al. 2006). Our goals were to determine (1) whether propagule availability and recruit composition vary between different groups of invasive and native bird-dispersed species (with a focus on canopy trees and shrubs), and among different habitats; and (2) whether dispersal and establishment of these species are influenced by various environmental factors, namely, distance to edge of forest patch, amount of herbaceous ground cover, leaf area index (LAI) and distance to nearest seed source.

Methods

Study region

Our study was conducted in the former ‘Big Scrub’ region of subtropical north-eastern New South Wales (NSW), Australia (28°68′–28°85′ S, 153°32′–153°44′ E). This region is situated on a basaltic plateau, ranging in altitude from 10 to 200 m A.S.L. with mean annual rainfall of 1,300 to 2,300 mm. Mean maximum daily temperatures range from 20–22°C in winter to 29–30°C in summer. Prior to European settlement, the area was covered by 75,000 ha of lowland subtropical rainforest, known as the Big Scrub, the largest tract of this forest type occurring in Australia (Floyd 1990). During the late 1800s, almost all of the original rainforest was cleared and now less than 1% of the original forest remains in the form of scattered remnants amongst farmland (Floyd 1990).

Patches of regrowth vegetation, usually dominated by invasive weeds such as bird-dispersed Cinnamomum camphora (camphor laurel) and Solanum mauritianum (wild tobacco) (Harden 1992–2002), are common on previously cleared land. Community land care groups actively undertake native vegetation restoration projects, which are usually relatively small areas comprising native rainforest species scattered throughout the region.

Replicate study sites

Sampling was conducted in three different habitats, each represented by three replicate sites (nine study sites in total). In order to maximise the coverage of the target population, one replicate site of each habitat type was located in the northern region of the Big Scrub area, one in the central region and a third in the south (Table 1). All sites occurred within a similar landscape context on previously cleared land, located at a similar distance (1–1.5 km) from the nearest significant native seed source (rainforest remnants). Habitats were:
Table 1

Outline of sampling design showing number of replicate trees of each canopy species (beneath which sampling stations were positioned) used at each site, number of seed traps and 1 m × 1 m quadrats per sampling station, and explanatory variables measured at each site

Habitat

Region

Canopy species

Replicate trees

Seed traps

Quadrats

Variables

Dist. to edge

LAI

Herb. cover

Tree regrowth

North

Cinnamomum camphora

12

3

2

Guioa semiglauca

12

3

2

Central

Cinnamomum camphora

12

3

2

Guioa semiglauca

12

3

2

South

Cinnamomum camphora

12

3

2

Guioa semiglauca

12

3

2

Native planting

North

Polyscias elegans

12

2

2

 

Alphitonia excelsa

11

2

2

 

Central

Omalanthus populifolius

11

2

2

 

Ficus coronata

11

2

2

 

South

Omalanthus populifolius

10

2

2

 

Macaranga tanarius

8

2

2

 

Shrub regrowth

North

Solanum mauritianum

12

2

2

 

Central

Solanum mauritianum

10

2

2

 

South

Solanum mauritianum

11

2

2

 

  1. (1)

    Tree regrowth”: Patches of forest each approximately 2 ha in area, which have established unassisted over the last 30–40 years on previously cleared land and contain a mixture of invasive and native plant species. Particularly common were the bird-dispersed trees, invasive C. camphora and native, Guioa semiglauca. This type of vegetation has been described in detail by Neilan et al. (2006).

     
  2. (2)

    Native plantings”: Restoration plantings comprising a mixture of native rainforest species. Despite some between-site variation in species composition, we considered these sites to be appropriate replicates since all contained a high proportion of fleshy fruited species, all were <1 ha in area and were planted 7–11 years prior to our study. This is supported by earlier studies suggesting that vegetation structure (Tucker and Murphy 1997) and distance from nearest seed source (White et al. 2004) are more important than species composition in influencing colonisation in native plantings.

     
  3. (3)

    Shrub regrowth”: Weedy regrowth containing invasive and native species, largely shrubs and herbaceous species, which typically occur in disturbed areas such as roadsides, creek banks and forest edges. All sites were <1 ha in area, linear in shape and dominated by invasive bird-dispersed shrub S. mauritianum.

     

Sampling stations

Sampling stations were established beneath the dominant fleshy-fruited canopy species at each site. Owing to differences in patch dimensions, the number of stations varied between habitats.

In each tree regrowth site, 24 sampling stations were established, 12 beneath the dominant invasive bird-dispersed tree C. camphora, and 12 beneath the most common native bird-dispersed tree G. semiglauca. In native plantings, the number of stations varied depending on the size of the site, and stations were distributed beneath the two most abundant native bird-dispersed species at each site (see Table 1). In each shrub regrowth site, 10–12 stations were established beneath S. mauritianum plants.

Distance to patch edge can influence seed deposition by frugivores (Buckley et al. 2006). Within tree regrowth sites, we measured distance from each station to patch edge. We did not record distance to edge in native planting and shrub regrowth sites since these were smaller; the ‘edge’ was sometimes difficult to define precisely and varied little between sampling stations.

At each sampling station, we measured LAI using a Li-Cor LAI-2000 Plant Canopy Analyser, taking a series of four readings at 0.5 m above ground level. Open-sky readings were obtained by taking the sensor into a nearby open area at the beginning and end of each measurement series for each station.

Soil seed bank germination

Soil samples were collected from each sampling station in February 2007 by removing coarse surface organic matter, then collecting a 250 × 250 × 80 mm (5,000 cm3) soil sample, to 8 cm deep from the A-horizon. Samples were spread in individual seed trays lined with paper towelling and placed in a greenhouse where they were irrigated twice daily. Emerging seedlings were identified, counted and removed at four-week intervals over 6 months.

Seed rain sampling

Seed traps were constructed from rectangular plastic trays (370 × 490 mm) with drainage holes in the base. Weed mat was taped into the base to prevent seed loss from the drainage holes. A removable piece of nylon fly mesh was placed on top of the weed mat to facilitate sample removal. Each trap was covered with a lid made of wire mesh with 12-mm-diameter holes, attached to a bamboo frame, three sides of which were fastened to the tray with plastic zip ties. The fourth side was attached with reusable rubber ties to allow easy access.

In tree regrowth sites, three traps were positioned beneath the canopy tree at each sampling station. In native planting and shrub regrowth sites, two traps were used per station. Where possible, we placed traps in positions where there was no (or very little) canopy overlap with neighbouring bird-dispersed plants.

Seed traps were emptied monthly between March 2007 and February 2008. All seeds and leaf debris were transported to the lab in paper bags and then sorted by sieving. Seeds with flesh or arils removed were presumed to have been dispersed by frugivores (Bleher and Böhning-Gaese 2001; White and Stiles 1992). These seeds were counted and identified using an existing seed herbarium as well as seed specimens collected from local fruiting plants throughout the duration of the study. Specimens were germinated to confirm identification where necessary.

All records of Ficus species were grouped owing to the morphological similarity of the seeds; however, the majority of Ficus records were probably F. coronata, the dominant species present.

Recruit surveys

At each sampling station, we surveyed recruits (seedlings and saplings <3 m in height) within two randomly positioned 1 m × 1 m quadrats. All recruits were identified and counted and the percent of the quadrat area covered by herbaceous vegetation was recorded. Where field identification was not possible, we collected herbarium specimens or seedlings for later identification, or sought identification from local restoration practitioners with expert knowledge of the region’s flora. Species taxonomy and nomenclature follow Harden (1992, 1993, 2000, 2002).

Distance to seed source

We selected two common invasive species, Lantana camara and C. camphora, and one common native species, Alphitonia excelsa, to use as model species for investigating the influence of distance to nearest seed source on seed rain density. These species were selected because seed sources of each were common within or close to all study sites.

We searched within a 50 m radius of each sampling station to locate the nearest mature individual of each species, and recorded its distance from the station. When no mature individual could be found, we recorded the distance as >50 m.

Data analysis

We grouped species into one of the following species categories according to growth form and status (invasive or native): native canopy trees, invasive canopy trees, native shrubs or understorey trees (henceforth ‘native shrubs’), invasive shrubs or understorey trees (henceforth ‘invasive shrubs’), native vines, invasive vines, native herbs and invasive herbs. For the in-depth analyses, we focussed on the two dominant growth forms: canopy trees and shrubs. Since our sampling areas for seed rain, seed bank germinants and recruits varied in size, we standardised all data to represent density (or species richness) 0.25 m−2. Seed rain data represent seeds 0.25 m−2 year−1.

Owing to the low species richness of invasive species recruits and seed rain, and for all species groups (native and invasive) in the soil seed bank, the ‘species richness’ data for these groups were not very informative. Therefore, while we present ‘density’ data for all species groups, we only discuss ‘species richness’ data for native canopy tree and shrub seed rain and recruits.

Data analyses were performed in GenStat (2008). In order to homogenise variances, all ‘density’ data were log10(x + c) transformed, where x = the value, and c = half the smallest recorded value (c = 0.24, 0.50 and 0.06 for seed rain, seed bank germinants and recruits, respectively) (Yamamura 1999). Transformed data were used in all analyses of density data outlined below. Transformation was not required for ‘species richness’ data, which were approximately normally distributed.

Density of seed bank germinants, seed rain and recruits for the four species categories (native canopy trees, invasive canopy trees, native shrubs and invasive shrubs) were subjected to split-plot ANOVA, with the model of three regions (blocks) by three habitat types split for sampling stations. This adopts the more conservative statistical approach, with the residual mean square (with four degrees of freedom) in the region by habitat stratum being used to test habitat differences. Likewise, we used split-plot ANOVAs to test for between-habitat differences in species richness of native tree and shrub seed rain and recruits. Pair-wise protected least significant difference (LSD) tests were employed for post-hoc multiple comparisons.

Linear regression analyses were performed to determine the effects of LAI, distance to patch edge, and herbaceous cover on seed rain, seed bank and recruit density and species richness for the four species categories (invasive and native canopy trees and shrubs). Sampling stations were used as the experimental units for the relationships, as the observed values for the independent variables across these stations represented the ranges of interest. Habitat was included as a factor in these regression analyses, which allowed the formal testing of the interaction of habitat by the independent term. For the sake of simplicity, only significant (P < 0.05) regression results are presented here, but comprehensive data are available from the authors on request.

We also performed correlations between all combinations of the three independent variables, LAI, distance to edge and herbaceous cover, to determine the degrees of relationship.

Two-way ANOVAs, followed by pair-wise LSD tests, were used to examine the effect of distance to nearest seed source (as a categorical factor) and habitat on seed rain 0.25 m−2 year−1 for A. excelsa, C. camphora and L. camara.

Results

Patterns of dispersal and establishment

During a 12-month period across all habitats, the seed rain contained 93,132 seeds of invasive species and 26,261 native seeds, representing 14 invasive and 112 native species. Of these, 7 and 41% of the invasive and native seeds respectively belonged to different species than the canopy tree, and were therefore considered to be dispersed seeds. Throughout this paper, we use the term ‘conspecific seed’ to refer to seed rain belonging to the same species as the canopy tree.

In total, 1,320 invasive and 767 native bird-dispersed recruits were recorded, representing 12 invasive and 46 native species. We recorded 2,901 invasive and 1,628 native germinants from the soil seed bank samples, belonging to 11 invasive and 21 native bird-dispersed species.

In all habitats, more species were recorded in the seed rain than amongst the recruits (Fig. 1). Most invasive bird-dispersed species were common across all habitats, and more ‘rare’ native species were recorded amongst the seed rain and recruits in tree regrowth than in the other habitats (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs11258-009-9587-2/MediaObjects/11258_2009_9587_Fig1_HTML.gif
Fig. 1

Rank abundance plots showing seed rain and recruit species abundance distributions in different habitats. Black markers represent invasive species and white markers, native species. Invasive species name abbreviations: Canopy trees: cc—Cinnamomum camphora; ll—Ligustrum lucidum; Shrubs: lc—Lantana camara; sm—Solanum mauritianum; ls—Ligustrum sinense; os—Ochna serrulata; cs—Cassia sp.; ss—Sambucus sp.; cr—Citrus reticulata; Vines: ap—Asparagus plumosus; ps—Passiflora spp.; Herbs: po—Phytolacca octandra; sn—Solanum nigrum; rh—Rivinia humilis; sp—Solanum pseudocapsicum

Canopy trees

Seed rain

Seed rain of two invasive bird-dispersed trees, C. camphora and L. lucidum, was recorded in all habitats (Table 2). Invasive seed rain density was higher in the tree regrowth sites than in native planting and shrub regrowth sites (Table 3). Seed rain density decreased with distance to patch edge (Table 4). This weak but significant pattern was attributable to conspecific seeds (C. camphora seeds landing beneath C. camphora trees), as no relationship was evident when these seeds were excluded from the analysis. Native seed rain was dominated by seeds of early successional species in all habitats (Table 2). Density was highly variable within habitats, and there was no significant difference between habitats. Species richness, however, was higher in native plantings than other habitats (Table 3).
Table 2

The three most abundant species of native and invasive tree and shrub represented amongst seed bank (SB), seed rain (SR) and recruits (R) in three habitats

 

Tree regrowth

Native planting

Shrub regrowth

SB

SR

R

SB

SR

R

SB

SR

R

Canopy trees

    Invasive

Cinnamomum camphoraE

×

×

×

×

×

×

 

×

×

Ligustrum lucidumE

 

×

×

 

×

×

 

×

 

    Native

Acacia melanoxylonE

×

×

 

×

 

×

×

×

×

Ficus spp.E

×

  

×

×

×

×

 

×

Alphitonia excelsaE

×

×

 

×

×

  

×

 

Guioa semiglaucaE

 

×

×

  

×

   

Commersonia bartramiaE

      

×

×

 

Mallotus philippensisE

  

×

     

×

Arytera diastilisL

  

×

      

Polyscias murrayiiE

    

×

    

Shrubs

    Invasive

Solanum mauritianumE

×

×

 

×

×

×

×

×

×

Lantana camaraE

 

×

×

 

×

×

 

×

×

Ligustrum sinenseE

 

×

×

 

×

×

 

×

×

Physalis peruvianaE

      

×

  

Ochna serrulataE

  

×

      

    Native

Rubus rosifoliusE

×

  

×

 

×

×

 

×

Omalanthus populifoliusE

   

×

×

×

 

×

×

Aphananthe philippinensisL

 

×

×

  

×

 

×

 

Pipturus argenteusE

×

  

×

  

×

  

Trema tomentosaE

 

×

  

×

  

×

×

Pittosporum undulatumE

 

×

  

×

    

Macaranga tanariusE

×

     

×

  

Wilkiea huegelianaL

  

×

      

Streblus brunonianusL

  

×

      

An ‘×’ indicates that a species is amongst the most common three species for a particular habitat (absence of an ‘×’ does not necessarily mean the species is absent)

Letters in superscript: E pioneer/early secondary species, L Late secondary/mature phase species (as described by Kooyman 1996)

Table 3

Mean density (N) of invasive (Inv) and native (Nat) tree and shrub seed bank germinants, seed rain and recruits in three habitats, and results of ANOVA analyses

 

Mean density per 0.25 m2

Tree regrowth

Native planting

Shrub regrowth

Canopy trees

Seed bank

    Inv

N

0.16

0.01

<0.01

    Nat

N

1.84

2.82

0.11

Seed rain

    Inv

N

10.28 (2.77)a(a)

0.86b(b)

1.24b(ab)

    Nat

N

22.46 (9.81)

65.22 (31.60)

6.79

SR

1.79 (1.56)b(b)

3.30 (3.00)a(a)

1.55b(b)

Recruits

    Inv

N

0.17

0.07

0.01

    Nat

N

0.41a

0.05b

0.02b

SR

0.23a

0.07b

0.02b

Shrubs

Seed bank

    Inv

N

0.56

0.63

7.67

    Nat

N

1.44

2.27

0.74

Seed rain

    Inv

N

3.24b(a)

9.20b(a)

1174.66 (16.23)a(a)

    Nat

N

1.59b(a)

5.63 (2.44)a(a)

0.38b(a)

SR

0.60

1.44 (1.20)

0.49

Recruits

    Inv

N

0.02

0.02

0.18

    Nat

N

0.04

0.06

0.05

SR

0.05

0.06

0.04

Mean species richness (SR) is shown for native species seed rain and recruits only. Values in parentheses represent density for dispersed seed only (conspecific seed excluded)

Letters in parentheses relate to seed rain means for dispersed seed only

Note: Means within rows with a common superscript are not significantly (P < 0.05) different from one another according to pair-wise protected least significant difference tests

Table 4

Results of linear regression analyses showing the effect of LAI, percent herbaceous cover and distance to patch edge (tree regrowth habitat only) on seed rain, seed bank and recruit density for invasive and native canopy trees and shrubs. Only significant results are presented. ‘Rel’ = Direction of relationship between explanatory and response variables; either positive (+), negative (−) or ‘variable’ (v), where a significant interaction exists

 

Overall effect (includes effect of ‘habitat’)

Effect of explanatory variable

Interaction

Response variable (in italics) & explanatory variable

P

F

Adj R2

P

F

Rel.

P

F

Invasive tree seed rain density (including conspecific seed)

Distance to edge (tree regrowth)

*

7.45

0.08

 

NA

  

Native tree seed bank density

LAI

**

8.80

0.15

*

4.32

+

  

Herbaceous cover

**

17.84

0.25

**

35.37

Invasive tree recruit density

Herbaceous cover

**

10.05

0.15

**

19.28

  

Distance to edge (tree regrowth)

**

14.51

0.18

 

NA

Native tree recruit density

LAI

**

29.59

0.45

**

36.29

+

  

Herbaceous cover

**

38.35

0.43

**

20.76

Native tree recruit species richness

LAI

**

26.04

0.37

**

41.31

+

  

Herbaceous cover

**

31.19

0.38

**

18.87

Invasive shrub seed rain density

Distance to edge (tree regrowth)

*

7.44

0.08

 

NA

+

  

Native shrub seed rain density

Distance to edge (tree regrowth)

**

31.31

0.30

 

NA

  

Native shrub seed rain species richness

Distance to edge (tree regrowth)

**

15.31

0.17

 

NA

  

Invasive shrub seed bank density

Herbaceous cover

**

14.81

0.21

**

28.68

+

  

LAI

**

47.50

0.51

*

6.05

+

Native shrub seed bank density

Herbaceous cover

**

10.75

0.07

**

10.75

  

Invasive shrub recruit density

 

Herbaceous cover

**

8.69

0.20

*

6.02

v

*

5.42

Distance to edge (tree regrowth)

**

11.20

0.15

 

NA

+

  

P < 0.05; ** P < 0.01

Seed bank

Invasive tree seed bank density was similar among habitats. Seeds of C. camphora were recorded in the seed bank in tree regrowth and native plantings, but L. lucidum seeds were not recorded in all habitats (Table 2). In all habitats, the native tree seed bank was dominated by seeds of early successional tree species (Table 1) and density was similar among habitats (Table 3).

Recruits

A similar density of invasive tree recruits was recorded among habitats (Table 3) and these were dominated by C. camphora. Ligustrum lucidum recruits were recorded in tree regrowth and native plantings, but not in shrub regrowth (Table 2). Invasive tree recruit density decreased with herbaceous cover in all habitats, and with distance-to-edge (in tree regrowth sites) (Table 4). Native recruits were dominated by early successional species, with the exception of Arytera diastilis, a later successional tree, only common in tree regrowth habitat (Table 2). A significantly higher density and species richness of native tree recruits was recorded in tree regrowth than other habitats (Table 3). Native recruit density and species richness were positively related to LAI and negatively related to herbaceous cover (Table 4).

Shrubs

Seed rain

Invasive shrub seed rain was dominated by S. mauritianum, L. camara and L. sinense in all habitats. Seed rain was higher in shrub regrowth than in other habitats, owing to the high density of conspecific seed. Dispersed seed rain density was similar between habitats (Table 3). In tree regrowth, invasive seed rain density had a weak positive relationship with distance-to-edge (Table 4). Native seed rain was dominated by early successional species and density was higher in native plantings than other habitats; again this was due to conspecific seeds. Native shrub seed rain density and species richness declined with distance-to-edge in tree regrowth (Table 4).

Seed bank

Density of invasive shrub seed bank germinants was highly variable and did not differ with habitat. Density increased with LAI and herbaceous cover (Table 4). Native shrub seed bank density was similar across habitats and negatively related to herbaceous cover (Table 4).

Recruits

Invasive shrub recruit density was similar among habitats (Table 3) and showed a positive relationship with distance-to-edge in tree regrowth vegetation. Response to herbaceous cover varied between habitats (Table 4); recruits increased with herbaceous cover in tree regrowth, but the relationship was negative in other habitats. Late successional native shrubs were common recruits in tree regrowth sites (Table 2), whilst early successional species dominated the other habitats. Recruit density and species richness did not differ among habitats.

Environmental variables

Herbaceous cover was negatively correlated with LAI (R = −0.22; P = 0.01). Neither LAI nor herbaceous cover was related to distance to patch edge in tree regrowth vegetation.

Distance to seed source

Seed rain density varied with distance to nearest seed source (F4,141 = 7.97; P < 0.01; F6,135 = 27.93; P < 0.01 and F6,140 = 17.81; P < 0.01 for L. camara, A. excelsa and C. camphora, respectively) (Fig. 2) and with habitat (F2,141 = 7.18; P < 0.01, F2,135 = 14.67; P < 0.01 and F2,140 = 5.26; P = 0.01 for L. camara, A. excelsa and C. camphora, respectively). Lantana camara and A. excelsa seed rain was lower in tree regrowth, and C. camphora seed rain higher in tree regrowth than in other habitats.
https://static-content.springer.com/image/art%3A10.1007%2Fs11258-009-9587-2/MediaObjects/11258_2009_9587_Fig2_HTML.gif
Fig. 2

Mean annual seed rain (±1 SE) at varying distance from nearest neighbour in three vegetation types for: a Invasive shrub Lantana camara. b Native tree Alphitonia excelsa. c Invasive tree Cinnamomum camphora

In shrub regrowth and native plantings, seed rain of both A. excelsa and L. camara occurred at significantly higher densities in seed traps positioned directly beneath the canopy of a mature individual than in traps positioned 1–10 m from the nearest seed source. In tree regrowth, A. excelsa and L. camara seed rain density at 1–10 m distance was similar to density beneath the parent plant, but declined significantly at 10–20 m from the nearest seed source.

Cinnamomum camphora trees were not present within native plantings; therefore, all sampling stations in this habitat were located at least 10 m from the nearest seed source. Cinnamomum camphora seed rain was higher at 10–20 m from the nearest seed source than at greater distances, but was similar among other distance categories (20–30 m to >50 m). Because tree regrowth sites were dominated by C. camphora, all sampling stations in this habitat were within 20 m of a seed source. Seed rain density was significantly higher directly beneath C. camphora trees than at greater distances, but was similar at 1–10 and 10–20 m from the nearest seed source. Cinnamomum camphora seed rain densities in shrub regrowth were highly variable and did not differ significantly with distance from seed source.

Discussion

Tree regrowth

Invasive tree seed rain within tree regrowth sites was somewhat higher near patch edges than the interior. This pattern was driven by C. camphora seed. Many species may produce more fruits near edges or in open sites than in the interior of patches (Restrepo et al. 1999; Bartuszevige and Gorchov 2006) and greater levels of frugivore activity may occur near edges (Malmborg and Willson 1988). These factors are likely to contribute to the observed pattern. In addition to greater seed input, higher densities of invasive tree recruits were recorded near edges. Distance to edge had a stronger effect on recruits than on seed rain, indicating that post-dispersal processes are probably contributing to this pattern. The intolerance of C. camphora to heavy shade (Dunphy 1988) may contribute to the higher recruit densities on patch edges.

Native tree seed rain was highly variable within habitats, and did not differ among habitats, whilst species richness (0.25 m−2 year−1) was higher in native plantings than tree regrowth. Recruits did not reflect this pattern; both density and species richness were higher in tree regrowth than native plantings. In addition, more ‘rare’ native species seeds and recruits were found in tree regrowth sites than other habitats, and although early successional recruits dominated tree regrowth, several late successional/mature phase species were also common. Previous studies have highlighted the potential value of C. camphora-dominated tree regrowth in facilitating establishment of native rainforest species (Kanowski et al. 2008; Neilan et al. 2006). Similarly, our results indicate that this habitat appears to provide a more suitable microenvironment for native tree recruitment than the other habitats.

Invasive shrubs, particularly L. camara, O. serrulata and L. sinense, were common in tree regrowth vegetation. Although S. mauritianum seeds were abundant in the seed rain and seed bank, recruits were relatively uncommon, probably owing to the poor ability of this species to establish under shade (Witkowski and Garner 2008). Invasive shrub recruit density was higher in the interior of tree regrowth patches than near the edge, and seed rain showed a similar (though weaker) pattern. This is likely attributable to L. sinense, the most common invasive shrub recruit in tree regrowth, which is capable of establishing beneath forest canopies (Morris et al. 2002).

Native shrub seed rain showed a different pattern to that of exotic shrubs, being higher near patch edges. Similar patterns of shrub seed rain occur in other forest types (Walker and Neris 1993). The higher seed rain near the edges of our study sites is probably partly due to the abundance of P. undulatum seed. Mature, fruiting individuals of this species are more common on forest edges (Rose 1997). The observed relationship is not entirely attributable to one species however, since species richness, as well as density, was higher at the edge.

Native shrub recruits and seed banks did not share this relationship with distance to edge. The discrepancy between seed rain and recruit distribution is probably because two of the three native shrub species that dominated the seed rain were early successional species that are common near forest edges, whereas the dominant recruits were later successional species that are less likely to occur near edges (Table 2).

The native shrub species dominating the seed bank also differed from the most abundant species represented amongst recruits and seed rain (Table 2). For instance, Pipturus argenteus and Rubus rosifolius were common in the seed bank, but recruits and seed rain of these species were uncommon. Pipturus argenteus seeds are long-lived in the seed bank (Enright 1985) and both P. argenteus and R. rosifolius germinate in large numbers from seed banks when placed in a high light environment, or when a gap is formed (Loh and Daehler 2007; Rogers 2000). The abundance of late successional recruits despite the dominance of seeds of early successional light-demanding species provides evidence that tree regrowth sites are developing rainforest-like characteristics.

Native plantings

Minimising invasive tree seed rain is especially important in native plantings as this habitat seems to provide an ideal microenvironment for the establishment of these species, particularly C. camphora. Given that native plantings were dominated by native plants, and other habitats by invasive species, it is not surprising that these sites received lower invasive tree seed rain than did other habitats. Despite the lower seed rain, however, invasive tree recruit density in native plantings did not differ from that found in tree regrowth. Successful recruitment of invasive trees in native plantings may be due to biotic factors (e.g. reduced competition for light, nutrients and moisture) conferred by the lower herbaceous cover at these sites (compared to shrub regrowth), or other factors such as distance from the parent plant; various studies have shown that seedling performance improves with distance from parent plant owing to factors such as reduced levels of competition, disease or predation (e.g. McAlpine and Jesson 2008; Wenny 2000).

Although native tree seed rain density was similar, and species richness was higher, in native plantings,compared with tree regrowth, recruit density and species richness were lower. In short, while seeds of a variety of native species were arriving, recruitment of these species was disproportionately low. Similar results have been reported to slow restoration elsewhere (Zimmerman et al. 2000). This might be partly due to high rates of herbivory and seed predation commonly occurring in restored sites (Woodford 2000), but is likely also due to the younger age of these sites, compared to tree regrowth. Tree regrowth sites have been establishing unassisted on abandoned land for several decades, whereas the majority of restoration work in this region has been conducted over the last 10–15 years. There is no doubt that structural complexity develops with age (Kanowski et al. 2003) and greater complexity, and the accompanying increased environmental heterogeneity have been shown to be related to higher species diversity (Le Brocque and Buckney 2003; Lenière and Houle 2006). Germination and establishment of the many native species recorded in the tree regrowth sites are likely to be influenced by the range of microsite characteristics present in these more heterogeneous systems, encompassing variation in the light environment, soil characteristics, moisture availability and temperature (Grubb 1977; Vieira and Scariot 2006), all of which may vary between native plantings and tree regrowth patches.

Native tree recruit density and species richness increased with LAI, supporting the hypothesis that greater structural complexity facilitates native tree recruitment. White et al. 2004 found that the number of native species colonising a restored site 15 years following initial planting had almost doubled from that recorded during sampling conducted 5 years earlier. With time, it is likely that native tree recruits will further increase in abundance and species richness in the restored sites used in this study. Whilst an ideal study scenario would involve a comparison of tree regrowth sites with native plantings of a similar age, no suitable older native plantings existed within our study area.

Solanum mauritianum, L. sinense and L. camara were the most abundant invasive shrub species represented amongst seed rain and recruits in native plantings, but S. mauritianum was the only species to emerge from the soil seed bank. Ligustrum sinense has a short-lived seed bank (Panetta 2000), which might explain its absence. Lantana camara has relatively low germination rates (Gentle and Duggin 1997), and so seeds of this species may be present in the soil, but might have failed to germinate from our samples.

As we observed in tree regrowth sites, the dominant native shrub species in the seed rain were not necessarily the most common recruits or seed bank germinants. For instance, T. tomentosa was common in the seed rain in native plantings (as well as in other habitats), but was not one of the most frequently recorded recruits. This may be explained in part by the low germination rates of T. tomentosa in shaded understorey conditions, and its relatively poor persistence in the soil seed bank (Raich and Khoon 1990).

Shrub regrowth

Both invasive and native tree recruits decreased with herbaceous cover, and high levels of herbaceous cover were characteristic of shrub regrowth. Of particular interest is the apparent absence of L. lucidum recruits in shrub regrowth, despite arrival of seeds in these sites. Mazia et al. 2001 found that L. lucidum seed addition to pasture failed to produce seedlings and it seems likely that the dense herbaceous layer in our study sites has a similar effect.

The herbaceous ground layer in shrub regrowth no doubt inhibits germination or establishment of native as well as invasive tree recruits, resulting in lower recruitment in these sites compared with that recorded in tree regrowth. Native tree seed is arriving in shrub regrowth sites, suggesting that succession at these sites is currently limited by post-dispersal processes, rather than propagule availability. Removal of grasses in abandoned pasture has been demonstrated to enhance germination and survival of native species recruits (Zimmerman et al. 2000), and this technique is commonly used in restoration of grassy areas in subtropical Australia (Kooyman 1996). Our findings offer support for this practice, although clearly additional efforts will be required to control invasive tree recruits when herbaceous cover is removed.

Shrub regrowth sites received more invasive shrub seed rain than other habitats, but the higher seed rain was not reflected in higher densities of recruits or seed bank germinants. The absence of significant differences in recruit and seed bank density between habitats is partly due to high variability within habitats but may also be attributable to the inhibitory effect of herbaceous cover on invasive shrub recruits in shrub regrowth sites. The seeds of one late successional native shrub species, Aphananthe philippinensis, were common in the seed rain in shrub regrowth sites, but early successional native shrubs dominated the recruits. Of the three habitats sampled in our study, shrub regrowth probably provides the least suitable establishment conditions for late successional species.

Distance to seed source

All of our study sites were at least 1 km from the nearest old growth rainforest remnant, but were very close to invasive species seed sources, thus limiting opportunities for the input of a high diversity of native seeds, and increasing the chances of arrival of invasive seed. Previous studies have shown high rates of colonisation by frugivore-dispersed invasive trees in sites with nearby invasive seed sources (e.g. Panetta and Sparkes 2001). Although birds transport seeds over long distances (Price 2006; Weir and Corlett 2007), most seeds tend to fall within 25 m of the parent plant (Wenny 2000) and dispersal beyond 60 m is rare (Clark et al. 2005). Like previous studies, our results showed a significant decrease in seed rain within less than 50 m of a seed source. This highlights both the importance of eliminating invasive seed sources from the vicinity of restored sites, and the benefits of retaining nearby native seed sources and focusing restoration efforts in areas close to native seed sources where possible.

Conclusions and management implications

The discrepancy among seed rain, seed bank and recruit composition for some groups of species (particularly native shrubs) highlights the complexity of dispersal and recruitment processes. Various factors, including different seed longevity, germination and establishment requirements among species, and variation in seed production between years, no doubt contribute to these differences.

Despite these complicating factors, we believe that it is possible to make some generalisations. Owing to the availability of native propagules and recruits in tree regrowth patches, these sites have the potential to be converted to native-dominated species assemblages. Two methods of managing tree regrowth dominated by C. camphora (‘staged’ and ‘patch’ removal of C. camphora) have been described by Kanowski et al. 2008. Both involve killing C. camphora trees and leaving the dead stems in situ, and both have been demonstrated to result in high levels of native species recruitment from the seed bank (Kanowski et al. 2008). In addition to selective removal of invasive trees, addition of propagules (particularly seeds of late successional species) may be beneficial at isolated tree regrowth sites to prevent ‘stagnation’ of secondary regrowth (Kooyman 1996).

Control of bird-dispersed invasive plants close to restoration sites is likely to greatly reduce the seed rain arriving in the site. Although control of nearby large tracts of C. camphora may be unrealistic, it would be worthwhile targeting small clusters or individuals of C. camphora and L. lucidum located within 20–30 m of plantings. Control of invasive trees in open areas is particularly important because for many species, individuals in open areas produce the greatest quantities of seed (Bartuszevige and Gorchov 2006).

Manipulation of herbaceous ground cover in restored sites and shrub regrowth will have an impact on the suite of species that establish at a site. Retaining herbaceous cover may hinder establishment of invasive trees and germination of invasive shrub seeds. However, removal of herbaceous cover is probably necessary for the establishment of native species. Further experimental work would be valuable in refining such practices.

Since native tree recruit density is associated with LAI, planting species with structural and growth characteristics that result in rapid canopy closure—a practice that is already used in restoration (Kooyman 1996)—is probably an effective way to facilitate the establishment of native recruits, particularly tree species, without promoting invasive species establishment. An abundance of native species seeds are available in native plantings, and so probably with time and management interventions such as weed control, these patches may develop into more species-rich and structurally complex systems.

Whilst shrub regrowth did not receive the numbers of ‘rare’ native species seeds recorded in the other habitats (especially tree regrowth), it nevertheless has relatively large numbers of native seeds arriving. With appropriate management interventions, e.g. control of herbaceous vegetation and strategic planting of a few hardy native species, we believe that this habitat also has the potential to move towards a native-dominated species assemblage.

Acknowledgements

S. Heuston, S. Jackson, P. Parrington, J. Morris, J. Rankin, Rous Water, and NSW National Parks and Wildlife Service provided access to their land. Land care practitioners T. Roberts, D. Bailey, R. Woodford and L. Gander assisted with plant identification. W. Neilan, H. Bower and D. Rankin from Byron Shire Council and J. Kanowski and C. Catterall of Griffith University provided research advice. S. Harvey, J. McCarthy, A. Dimmock and M. Landos provided technical assistance. Thanks to D. Mayer for performing data analyses and D. Panetta and two anonymous reviewers for constructive comments on the manuscript. This study was funded by the Cooperative Research Centre for Australian Weed Management and Queensland Department of Primary Industries and Fisheries.

Copyright information

© Springer Science+Business Media B.V. 2009