Rain forest understorey ferns facilitate tree seedling survival under animal non-trophic stress

Journal of Vegetation Science 23 (2012) 847–857

Rain forest understorey ferns facilitate tree seedlingsurvival under animal non-trophic stress

Guo-Zhang Michael Song, David J. Yates & David Doley

Keywords

Australian rain forest; Facilitation; Litter stress;

Plant–plant interaction; Species trait;

Vertebrate disturbance

Nomenclature

Spencer et al. (1995) for ferns; Simpson & Day

(1999) for birds

Abbreviations

LSa = litter smothering due to abiotic

disturbance; LSb = litter smothering due to

bird disturbance; WR = wilting or rotting

Received 1 April 2011

Accepted 14 January 2012

Co-ordinating Editor: Kerry Woods

Song, G.-Z.M. (corresponding author,

[email protected]): Institute of Ecology

and Evolutionary Biology, National Taiwan

University, No. 1, Section 4, Roosevelt Rd,

Taipei, 10617, Taiwan

Yates, D.J. ([email protected]) & Doley, D.

([email protected]): Centre for Mined Land

Rehabilitation, The University of Queensland,

Brisbane, Qld, 4072, Australia

Abstract

Questions: Although forest ecosystems harbour many animal species and ani-

mal non-trophic effects are as ubiquitous as trophic effects, few studies have

examined animal non-trophic effects on plant–plant interactions. Can animal

non-trophic behaviour influence plant–plant interactions and thus, change the

net effect of interactions, especially those between understorey vegetation and

tree seedlings? How do the species traits of understorey vegetation and animals

contribute to their interactions with seedlings?

Location: TheMain Range National Park, southeast Queensland, Australia.

Methods: Seedling survival related to the cover of understorey vegetation

(mainly the fern Lastreopsis decomposita) was monitored in a 0.5-m wide and

200-m long transect for 2 yrs. Ten per cent of the transect was caged to estimate

the effects of non-trophic disturbances from two pheasant-size, ground-dwelling

birds (Menura alberti and Alectura lathami) for 1 yr. Two hundred plastic tags

(5 9 5 cm)were placed in the transect to quantify bird disturbance and litter input.

Results: The negative effects of the fern understorey on seedlings were

increased light deficits, greater risk of herbivory and wilting or rotting; the posi-

tive effects were reduced uprooting and litter smothering due to abiotic and bird

disturbances. Notably, the exclusion of bird activity changed the net effect of the

fern understorey from positive to neutral, and seedling survival was positively

correlated with fern cover.

Conclusions: The net effect of plant-plant interactions is subject to change

when additional species are involved. In addition to trophic effects, non-trophic

effects are such that they can change interactions between plants. A conceptual

model of species traits contributing to interactions is presented.

Introduction

In forest ecosystems, the influences of understorey vegeta-

tion on the early regeneration of tree species can be as

important as those of overstorey trees (Royo & Carson

2006). Due to the profound light interception effect of un-

derstorey vegetation (Harms et al. 2004; Montgomery

2004), it may havemore influence than overstorey vegeta-

tion on the light environment of the forest floor (George &

Bazzaz 1999a,b; Montgomery 2004) and the establishment

of seedlings (George & Bazzaz 1999a,b; Harms et al. 2004).

Moreover, interactions between tree seedlings and the un-

derstorey vegetation are inevitable because the two com-

munities share the same microhabitat, and seedlings

generally spend a long time in the understorey (e.g. Harms

et al. 2004). Therefore, understorey vegetation may ulti-

mately shape the composition and structure of forests

through modifying seedling banks (Denslow et al. 1991;

George & Bazzaz 1999a,b).

The majority of previous studies have focused on the

pair-wise interactions between understorey vegetation

and seedlings (e.g. Royo & Carson 2006). The documented

effects of understorey vegetation on seedlings include

resource competition, litter stress and allelopathy. For

example, the growth rate of tree seedlings can be reduced

by understorey vegetation via light interception (e.g.

George & Bazzaz 1999a) or competition for nutrients and

water in the soil (e.g. Yamasaki et al. 2002). Understorey

vegetation can also damage or smother seedlings through

falling litter (e.g. Drake & Pratt 2001). Phytochemical

Journal of Vegetation ScienceDoi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science 847

compounds released into soil by understorey vegetation

may significantly reduce seedling growth and survival

through interfering with metabolism, nutrient uptake and

root development (e.g. Wardle et al. 1998; Walker et al.

1999). In general, the pair-wise interactions between un-

derstorey vegetation and seedlings are negative.

A few studies have examined more complex interactive

relationships amongst more than two species or guilds

(Rebollo et al. 2005; Orrock et al. 2010). In a Costa Rican

rain forest, Denslow et al. (1991) reported that understo-

rey palms and cyclanths posed an increased risk of patho-

gen attack and herbivory on tree seedlings by harbouring

pathogens and herbivores. This has been described as ref-

uge-mediated apparent competition (Orrock et al. 2010).

In such types of interaction, understorey vegetation indi-

rectly exerts a negative effect on seedlings through provid-

ing refuges for pathogens and herbivores. Interestingly,

refuge-mediated effects can also be positive, e.g. even in

the presence of competitive interactions, some unpalatable

plants protect neighbouring plants of other species from

intensive livestock grazing and thus facilitate their growth

and survival (Callaway et al. 2005; Rebollo et al. 2005;

Osem et al. 2007). Many plant–plant interactions appear

to be positive when effects of additional interacting species

are incorporated into the interactions (Levine 1999; Rebol-

lo et al. 2005; Brooker et al. 2008).

Whilst most reported effects of understorey vegetation

on seedlings are negative, many of these conclusions are

drawn from studies that only examined pair-wise plant–

plant interactions (e.g. Royo & Carson 2006). Forests har-

bour many kinds of animal, so the involvement of animals

in plant–plant interactions is hardly surprising. However,

few studies have examined animal effects on interactions

between understorey vegetation and tree seedlings.

Studies examining animal effects on plant–plant interac-

tions have focused on the effects of animals as consumers

through herbivory and predation (Brooker et al. 2008). In

fact, animals can also exert profound non-trophic, physical

effects on the growth and survival of plants, such as tram-

pling and burrowing (Ickes et al. 2001), or otherwise mod-

ifying the physicochemical environment. For example,

earthworms facilitate the establishment of legume seed-

lings through increasing soil phosphates (Thompson et al.

1993). Although there is growing recognition that non-

trophic effects have significant impacts on ecosystem pro-

cess and structure (Boogert et al. 2006; Wright & Jones

2006), their effects in mediating plant–plant interactions

have received little attention.

This study aims to examine the relationships between

understorey vegetation, tree seedlings and non-trophic

effects of animals. We conducted our experiment in an

Australian subtropical rain forest, focusing on non-trophic

disturbances on the forest floor caused mainly by two

pheasant-size, ground-dwelling birds, the Albert’s lyrebird

(Menura alberti) and the Australian brush-turkey (Alectura

lathami) (Song 2007). Although these two bird species feed

on invertebrates in soil and plant litter, their litter-raking

behaviours and the building of egg-incubating mounds by

the brush-turkey contribute to a high proportion of seed-

ling mortality through uprooting and litter smothering

(Song 2007). Both species avoid foraging where short-

stature understorey vegetation is present (Ashton & Bassett

1997). The understorey vegetation in our study site was

dominated by the trim shield fern (Lastreopsis decomposita).

By assessing the survivorship of tree seedlings under dif-

ferent levels of fern cover and in the presence or absence

of non-trophic bird disturbance, we evaluated the effects

for all elements of the interaction network and identified

relationships between species traits and these element

effects. We asked: (1) can the litter-raking disturbance

caused by birds change interactions between understorey

vegetation and seedlings; (2) what are the negative and

positive effects of understory vegetation on seedlings; and

(3) how do the species traits of this fern and the two birds

species contribute to their interactions with seedlings?

Methods

Study site

All fieldwork was conducted in a transect 0.5-m wide and

200-m long located on a gentle west-facing slope of a rain

forest stand in the Main Range National Park, south-

east Queensland, Australia (28°13′55″S, 152°25′23″E,1100 m a.s.l.). The climate at the site is cool and subtropical.

The average annual rainfall at the Gambubal Forest Station

(28°14′50″S, 152°23′07″E, 930 m a.s.l.) is 1263 mm; mean

monthly temperature ranges from 4.9 (July) to 25.2 °C(January), winter minimum temperatures falls below 0 °C,with frosts being common on open sites, and maximum

temperatures in summer exceed 30 °C. The study site is

occupied by a notophyll vine forest (NVF)with a closed can-

opy, about 35–40 m tall. Sloanea woollsii is the dominant

canopy tree, accounting for 54% of the total basal area,

while the most abundant canopy tree is Doryphora sassafras,

with 25% of the total stem number. Please refer to Penfold

& Lamb (1999) and Debski et al. (2000) for detailed descrip-

tions of the geology and vegetation of this site.

Seedling dynamics

Within the 200-m transect, every tree seedlingwith a height

equal to or shorter than 15 cm was tagged, mapped and

identified to species. Seedling attributes were recorded

monthly for 2 yrs from October 2002: height, number of

cotyledons, number of leaves, extent of herbivory and the

percentage of foliage covered by litter. When a seedling was

Journal of Vegetation Science848 Doi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science

Ferns, animal non-trophic effects and seedling survival G.-Z. M. Song et al.

dead or missing, the cause was inferred from an examina-

tion of the dead plant body, any trace of bird disturbance on

the forest floor and previous observations recorded for that

seedling (see more details in Song 2007). Agents of seedling

mortality were identified as: (1) uprooting, (2) litter smoth-

ering due to abiotic disturbance (LSa), (3) litter smothering

due to bird disturbance (LSb), (4) herbivory, (5) wilting or

rotting (WR) and (6) unknown agents.

A dead seedling found on the forest floor with its roots

exposed, was presumed to have been uprooted. LSa was

defined as seedling death due to litter smothering caused

by gravity (litter fall), wind, rain or microtopography. LSb

was defined as seedling death through litter smothering

from the foraging behaviour of Albert’s lyrebird and Aus-

tralian brush-turkey. A low degree of herbivory does not

pose an immediate risk for seedling survival (Blundell &

Peart 2001); therefore, herbivory was only recorded as the

cause of mortality if more than 80% of foliage was con-

sumed in the last census. Due to the difficulty in distin-

guishing seedling deaths caused by water stress from

deaths by pathogen attack, causes of the death of wilted or

rotted seedlings were all recorded as WR. If the condition

of the dead seedlings did notmeet any of the above criteria,

the cause of death was recorded as unknown.

Effects of the fern understorey on environmental factors

Understorey vegetation is defined here as all plant cover

less than 50-cm tall. Its cover was assessed at 1-m intervals

by visual estimation to the nearest 10%. Lastreopsis

accounted for more than 90% of the understorey vegeta-

tion cover over the 200-m transect.

Twenty-six locations with fern cover ranging from 0 to

90% were sampled along the 200-m transect to assess the

light transmittance of the fern crown. Light transmittance

was expressed as the ratio of the transmitted photon flux

density (%TPFD) at a height of 0.08 m (under the fern

crown) to that at a height of 1.0 m (above the fern crown)

under overcast sky conditions (Figure S1). The%TPFD

above or under the fern crown was expressed as a percent-

age of the above-tree canopy photon flux density (PFD)

assessed instantaneously. The PFD under or above the fern

crown was measured with a line quantum sensor and a

LI-250 light meter (Li-Cor, Lincoln, NE, US). Above-tree

canopy PFD was assessed with a LI-190 point quantum

sensor and a LI-1400 data logger (Li-Cor) placed on the top

of a 37-m high weather tower located ca. 100 m from the

transect. The PFD was assessed on 10 September 2005

when the sky was uniformly overcast.

Two types of tag were placed side by side at 1-m inter-

vals along the centre line of the 200-m transect to assess

the effects of the fern understorey on the frequency of bird

disturbance and the amount of litter input. These tags

(50 mm 9 50 mm 9 1 mm) were made from pieces of

dark green plastic sheet, which minimized the visual

attraction to birds and was heavy enough that it would not

be removed or overturned by wind. Type 1 tags were used

to measure the amount and cause of monthly litter input.

Every Type 1 tag was laid on the surface of the ground and

fastened down with a 10-cm long peg. The cover and

thickness of litter on each tag were recorded to derive the

net litter input. The cause of litter movement (abiotic or

bird disturbance) was inferred by examining whether

there were signs of litter-raking activity on the ground

within 30 cm of the tag. In order to precisely quantify the

periodic litter input, litter on tags was always completely

removed after the monthly observations. Tags were

returned to the ground surface if they were buried deeply

or had been dislodged by birds. Type 2 tags were designed

to detect the frequency and cause of litter removal (abiotic

or bird disturbance); they were not fixed to the ground

and were placed directly alongside Type 1 tags. The dis-

tance of Type 2 tag displacement and the causes of tag

movement (see more details in Song 2007) were recorded.

Displaced Type 2 tags were returned to their original loca-

tion after each measurement and missing tags were

replaced whenever necessary. Measurements were carried

out monthly fromNovember 2003 to January 2005.

The effects of fern crowns on bird disturbance and litter

input were examined in terms of the cover and the prox-

imity of fern crowns to tag sites. The extent of fern cover

for each tag was estimated visually at 1-m intervals along

the transect. The proximity of fern crowns to tags was cate-

gorized into three groups: ‘without fern overhead’ (at least

5 cm beyond vertical projection of fern crowns), ‘at the

edge of fern crown’ (within 5 cm of edge of fern crown

projection) and ‘under fern crown’ (more than 5 cm inside

fern crown projection).

Animal exclosure

The effect of bird non-trophic disturbances on the interac-

tion between understorey vegetation and tree seedlings

was examined using animal exclosures. These were con-

structed from chicken wire (1.5 cm 9 1.5 cm mesh size),

which prevented intrusion or damage by both chicks and

adults of the Albert’s lyrebird and Australian brush-turkey.

This mesh size was still large enough to allow passage of

most invertebrates (e.g. slugs and insects), which are

responsible for the majority of seedling herbivory (Song

2007). Wire cages of this mesh size have a negligible effect

on internal micro-environment (Bridgeland et al. 2010).

Ten cages (2-m long 9 0.7-m wide 9 1-m high) were

placed at 20-m intervals along the transect so that 10% of

the total area of the 200-m transect and 10% of both types

of tag were protected by cages. The tops of cages were left

Journal of Vegetation ScienceDoi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science 849

G.-Z. M. Song et al. Ferns, animal non-trophic effects and seedling survival

open to allow litter fall, including the macro-litter fall

(leaves and woody debris >30 cm 9 1.5 cm) reported to

be an important cause of seedling death (Clark & Clark

1991; Gillman et al. 2004). Therefore, the exclosure

mainly eliminated LSb and uprooting caused by the two

semi-flightless birds, whilst LSa could still be a possible

cause of seedling mortality. The exclosure experiment

commenced in September 2003 and lasted for 1 yr.

Data analysis

Litter amount

The litter volume V (cm3) on tags was obtained by multi-

plying the area of the plastic tag A (cm2), litter thickness

T (cm) and litter cover C (%) on tags (V = A*T*C). For easeof quantitative comparison of litter amount, instead of

litter volume, in the present study litter amount was

expressed as average depth D (D = V/A). For example, if

litter 1-cm thick covered 20% of the area of a tag, the litter

amount on this tag was expressed as 0.2-cm deep.

Pseudo-replication and spatial autocorrelation

Field studies carried out on continuous transects or plots

are subject to pseudo-replication (Hurlbert 1984). To test

whether our environmental data collected from adjacent

sampling points were independent from each other, spatial

autocorrelation was examined with semivariance analysis

(Palmer 1988). The examined environmental factors, annual

frequency of bird disturbance (detected by 180 uncaged

Type 1 and 180 uncaged Type 2 tags), annual amount of

litter input (uncaged Type 1 tags) and fern cover, were

compared over sampling intervals between 1 and 50 m

(Fig. 1). Semivariances for annual frequency of bird distur-

bance (Fig. 1a) and annual litter input (Fig. 1b) changed

little, indicating no evidence of autocorrelation for bird

disturbance or litter input at sampling intervals as close as

1 m. However, for fern cover, the upward trend of the

semivariance curve between sampling intervals of 1 m and

3 m implies spatial autocorrelation at <3 m (Fig. 1c).

Therefore, to eliminate spatial autocorrelation in regres-

sion analyses between fern cover and other environmental

factors, fern cover and all other measured environmental

factors were calculated as the running means of 5-m

sections over the 200-m transect.

Effects of fern cover on seedling survival

The average hazard rate (total number of seedling deaths

divided by the sum of observed survival times) was used to

quantify the probability of seedling survival (Kleinbaum

1996). A higher average hazard rate indicates a lower

probability of survival. The effects of fern cover on seedling

survival was estimate with the Cox proportional hazards

model, a semi-parametric regression method of survival

analysis (Kleinbaum 1996). In addition to fern cover, eight

other predictor variables (covariates) that might affect seed-

ling survival were included in our analysis (Appendix S1):

three properties of habitats (frequency of bird disturbance,

PFD above fern crown and litter thickness), two environ-

mental stresses on seedlings (herbivory and litter cover

observed on seedlings), and three properties of seedlings

(species, age and seedling initial height). Eight species with

more than ten seedlings were included in survival analysis

(see Appendix S1 for species names). Forward step-wise

Distance of sampling interval (m)1 10 100

100

1000

Sem

ivar

ianc

e

1

10

100

1

10

100 (a)

(b)

(c)

Fig. 1. The spatial autocorrelation of (a) annual frequency of disturbance

(yr�1) of two ground-dwelling birds (Menura alberti and Alectura lathami)

detected using 180 uncaged Type 1 and 180 uncaged Type 2 tags, (b)

annual litter input (cm·yr�1) detected with 180 uncaged Type 1 tags, and

(c) cover (%) of the fern understorey (Lastreopsis decomposita) over the

200-m transect. All curves are parallel to the x-axis, indicating no evidence

of autocorrelation for adjacent sites over the whole range (1–50 m),

except fern cover (spatial autocorrelation up to 3 m). Note log scale used

on x- and y-axes.

Journal of Vegetation Science850 Doi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science

Ferns, animal non-trophic effects and seedling survival G.-Z. M. Song et al.

selectionwas employed to identify factors that significantly

influenced seedling survival. Survival analysis was per-

formed with SPSS software (version 12.0; SPSS Inc., Chi-

cago, IL, US).

Effects of fern cover on seedling mortality agents

The effects of fern cover on seedling mortality agents

were examined by analysing the contribution of each

agent to seedling mortality under different levels of fern

cover (Appendix S2) with principal components analysis

(PCA; performed with R program, version 2.12; R Foun-

dation for Statistical Computing, Vienna, Austria). In

PCA, levels of fern cover and contribution of seedling

mortality agents were treated, respectively, as the explan-

atory and response variables. The Euclidean distance cal-

culated in PCA gives too much weight to null and small

values, making standard PCA unsuitable for analysis of

data with null or small values (Legendre & Gallagher

2001). To reduce the undesired Euclidean distance

effects, data were pre-processed with the Hellinger dis-

tance transformation before PCA was performed (Legen-

dre & Gallagher 2001).

Results

Seedling dynamics

From September 2002 to October 2004, a total of 1893

seedlings, representing 23 tree species, were recorded in

the 200-m transect. During this time, 1461 seedlings died

(including five snapped off by the observer) and 432

remained alive at the end, and the seedling population size

at any time ranged from 275 to 917. Of the 1456 seedlings

that died from natural causes, herbivory caused the largest

proportion of seedling mortality (28.2%), followed by LSa

(27.8%), WR (18.8%), uprooting (15.4%), LSb (7.9%)

and unknown agents (1.9%). Litter-related disturbances

(LSa, LSb and uprooting) were responsible for more than

half (51.1%) of seedling mortality.

Abiotic and bird disturbance

In the 2160 measurements (180 uncaged Type 1 and 2 tag

sites 9 12 mo) on the seedling transect, a total of 1755 dis-

turbances were recorded. Of these, 826 (47.07%) were

caused by birds, 767 (43.70%) by abiotic agents and 162

(9.23%) by both agents. Only four abiotic disturbanceswere

caused by branches thicker than 1 cm. In other words, the

majority of litter objects causing LSa were composed of dead

leaves and small debris. This is consistent with studies of

Hegarty (1991) and Jackson & Bach (1999), indicating that,

instead of the physical damage reported by Clark & Clark

(1991), litter in this Australian rain forest imposed stress on

seedling survival chiefly through smothering.

Effects of the fern understorey on environmental factors

The fern understorey markedly reduced light availability,

bird disturbance and litter input. Light transmission through

the fern crown decreased as fern cover increased (Fig. 2),

suggesting that light interception is one of the negative

effects of ferns on seedlings. Positive effects of the fern un-

derstoreywere also identified, including reduced bird distur-

bance and litter stress. The annual frequency of bird

disturbance and the annual amount of litter input decreased

exponentially and significantly with increasing fern cover

(Fig. 3a, b). Frequency of bird disturbance and the amount

of litter input were significantly lower at the edge of or

under fern crowns thanwhere the fernwas absent (Fig. 4).

Effects of the fern understorey onmortality agents

The presence of a fern understorey affected relative contri-

butions of mortality agents to seedling death. Sites with

fern cover of 0–30% are separated from those with fern

cover over 30% along the first PCA axis (Fig. 5). LSb is the

response variable most closely and positively related to the

first PCA axis, and WR is most closely and negatively

related to this axis (Fig. 5). The contribution of LSb to

seedling mortality decreased and the contribution of WR

increased with increasing fern cover. Sites with fern cover

31–70% and 71–90% were further separated on the sec-

ond PCA axis. The second PCA axis was positively related

to LSa and uprooting and negatively related to herbivory

(Fig. 5). The stress of LSa and uprooting can be reduced

further at the highest fern cover, although the proportion

of seedling mortality caused by herbivory increases.

Effects of animal exclosure

Cages reduced the frequency of bird disturbance but not

the frequency of abiotic disturbance. Annual frequencies

of bird disturbance detected by Type 1 and 2 tags in

cages (0.50 ± 0.14 yr�1) were significantly lower than

for uncaged tags (5.49 ± 0.22 yr�1) (ANOVA, F = 57.38,

P < 0.001). In contrast, there was no significant differ-

ence between the averages of annual frequency of abiotic

disturbance within cages (4.70 ± 0.51 yr�1) and outside

cages (5.16 ± 0.19 yr�1) (ANOVA, F = 0.62, P = 0.43).

The net effects of the fern understorey on seedling

survival

The average hazard rate of uncaged seedlings decreased

with the increase of fern cover, indicating a positive net

Journal of Vegetation ScienceDoi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science 851

G.-Z. M. Song et al. Ferns, animal non-trophic effects and seedling survival

effect of ferns on seedlings (Fig. 3c). The survivorship of

uncaged seedlings increased as fern cover increased

(regression coefficient of the Cox proportional hazards

model = 0.994, P < 0.001). In contrast, when seedlings

were caged, the fern understorey had no significant

influence (neither positive nor negative) on seedling

survival (P = 0.733), i.e. the exclusion of bird distur-

bance turned the net effect of this fern understorey on

seedling survival from positive to neutral. In addition to

fern cover, six other factors (frequency of bird distur-

bance, herbivory on seedling, litter cover on seedling,

species, age and initial height) also had significant influ-

ences on the survival of caged or/and uncaged seedlings

(Appendix S1).

Discussion

Effects of the fern understorey on tree seedlings

In this subtropical rain forest, the fern understorey reduced

the main environmental stress, litter-related disturbances

(LSa,LSbanduprooting), and in turn facilitated the survival

of tree seedlings. Facilitation tends to be most commonly

observed in habitats with relatively high abiotic or biotic

environmental stresses (Levine 1999; Brooker et al. 2008),

and the likelihood of positive interactions between plants

increaseswith the severityof environmental stresses (Maes-

tre et al. 2005; Kikvidze et al. 2006; Brooker et al. 2008).

One speciesmay facilitate the establishmentof other species

solely through reducing abiotic or biotic environmental

stresses on the facilitated species (Levine 1999). For exam-

ple, in a mixed temperate Patagonian forest where desicca-

tion stress was responsible for most seedling mortality of

Nothofagus dombeyi in open areas, a bamboo understorey

(Chusquea culeou) facilitated Nothofagus early survival by

reducing the rate of soil drying (Caccia et al. 2009). In our

study site, litter-related disturbances were one of the major

environmental stresses, being responsible for 51% of seed-

ling mortality. The fern understorey reduced bird distur-

bance and litter input (Figs. 3a, b, 4), and consequently

reduced the risk of seedlings dying from LSa, LSb or

uprooting (Fig. 5), and ultimately facilitated seedling

survival (Fig. 3c).

Effects of animal non-trophic effects on plant–plant

interactions

Out results indicated that, in addition to trophic effects,

non-trophic effects of animals could be significant forces

Fern cover (%)0 20 40 60 80 100

Ligh

t tra

nsm

ittan

ce ra

te (%

)

10

100

1000

y = 91.60e–0.011x

r2 = 0.55; P < 0.001

Fig. 2. The light interception effect of the fern understorey (number of

sample sites = 26). Light transmittance rate of the fern understorey

decreases with the increase of fern cover. In sites where fern cover was

low, due to the complexity of light environment in the understorey, the

transmittance PFD above the fern crown (1 m above the ground) was

lower than that under the fern crown (8 cm above the ground), which

contributed to the light transmittance rate >100%. Note log scale used on

y-axis.

(c)

(b)

(a)

Fern cover (%)0 10 20 30 40 50 60

Aver

age

haza

rd ra

teof

unc

aged

see

dlin

gs(d

eath

mon

–1)

0.01

0.1

1

Am

out o

f litt

er in

put

(cm

yr–1

)

1

10

100

Freq

uenc

y of

bird

dist

urba

nce

(yr–1

)

0.1

1

10

100y = 6.33e–0.017x

r 2 = 0.28; P < 0.001

y = 7.96e–0.013x

r 2 = 0.40; P < 0.001

y = 0.25e–0.017x

r 2 = 0.35; P < 0.001

Fig. 3. The relationship between fern cover (number of sites = 40) and (a)

frequency of bird disturbance, (b) amount of litter input and (c) average

hazard rate of uncaged seedlings. The fern cover is the running means of

5-m sections over the 200-m transect. Bird disturbance, litter input and

average hazard rate decrease exponentially as fern cover increased. Note

log scale used on y-axis.

Journal of Vegetation Science852 Doi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science

Ferns, animal non-trophic effects and seedling survival G.-Z. M. Song et al.

influencing the net outcome of plant–plant interactions.

The contemporary core theories of biotic interactions in

ecology are based mainly on trophic interactions (van

Veen et al. 2009), so that the food web is an essential

concept in ecology textbooks (e.g. Krebs 2001). Like-

wise, studies reporting facilitative interactions between

plants have concluded that facilitators increase the sur-

vival or growth of facilitated species by reducing the tro-

phic effects of herbivores (e.g. Brooker et al. 2008).

However, non-trophic effects of animals appear to be

present in many ecosystems and their impacts can also

significantly shape ecosystem process, structure, function

and biodiversity (Boogert et al. 2006; Wright & Jones

2006). In some situations, herbivores may even influ-

ence plant communities more in non-trophic than in

trophic ways (Wilby et al. 2001). Overall, our results

highlight the need to incorporate non-trophic effects in

the core theories of community ecology.

Many studies have examined only pair-wise interac-

tions between understorey vegetation and tree seedlings,

and the involvement of a third factor has not been consid-

ered (e.g. Royo & Carson 2006). In the present study, the

net effect of plant–plant interactions changed with the

extent of animal disturbances, reflecting the significant

effects of animals on interactions between plants. Forest

ecosystems are rich in faunal diversity and abundance, so

without the incorporation of animal effects, we can hardly

obtain a comprehensive understanding of plant–plant

interactions. Our study highlights the importance of incor-

porating animal effects into the interactions between plant

communities in general.

Contributory species traits andmechanisms of

interactions

The occurrence of certain types of interaction is associated

with species-specific traits or certain assemblages of species

traits (Levine 1999; Brooker et al. 2008; van Veen et al.

2009); hence, trait identification for the interacting species

will promote understanding of interactions (van Veen

et al. 2009). The net effect of Lastreopsis is neutral on caged

seedlings but positive on uncaged seedlings. These findings

are contrary to those of most previous studies. Here, we

propose a conceptual model to elucidate the interaction

mechanisms through identifying the contributory species

traits of Lastreopsis (marcescence, dense foliage arrange-

ment and short stature) and the two ground-dwelling birds

(small body size and non-folivorous diet) (Fig. 6).

Species traits of the fern

The neutral net effect of Lastreopsis on caged seedlings can

be attributed to its traits of marcescence and dense foliage

Freq

uenc

y of

bird

di

stur

banc

e (y

r–1)

0

2

4

6

8Am

ount

of l

itter

inpu

t(c

m y

r–1)

0

2

4

6

8

10

12

A

(a)

(b)

A

B

A A

B

Under ferncrown

At the edge of fern crown

Without fernoverhead

Location of tag

Fig. 4. The effect of the proximity of the fern understorey on (a)

frequency of bird disturbance (mean ± SE) and (b) amount of litter input

(mean ± SE). Significant differences among categories (P < 0.05) are

indicated with different letters above bars. ‘Under fern crown’ are uncaged

tags (n = 17) located in areas at least 5 cm inside the projected fern crown

on the forest floor; ‘At the edge of fern crown’ are uncaged tags (n = 64)

in areas 5 cm inside or outside the projected fern crown; and ‘Without

fern overhead’ are uncaged tags (n = 99) in areas at least 5 cm away from

the projected fern crown.

PCA axis 1–2 –1 0 1 2

PC

A a

xis

2

–2

–1

0

1

Mortality agentFern cover

0-30 %Fern cover

71-90%Fern cover

31-70%Fern cover

Unknown

WR

LSa

Herbivory

UprootingLSb

Fig. 5. Principle components analysis (PCA) ordination diagram

illustrating the effect of fern understorey on seedling mortality agents.

Explanatory variables (different levels of fern cover) appear as dots and

response variables (contribution of seedling mortality agents) appear as

vectors. LSa, litter smothering due to abiotic disturbance; LSb, litter

smothering due to bird disturbance; WR, wilting or rotting.

Journal of Vegetation ScienceDoi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science 853

G.-Z. M. Song et al. Ferns, animal non-trophic effects and seedling survival

arrangement. Marcescence – retention of senescing foliage –

reduces themechanical damage to seedlings from litter fall.

Litter fall damage contributed from understorey vegetation

may be responsible for much of the seedling mortality in a

forest (e.g. Gillman & Ogden 2005). In a Hawaiian rain

forest, up to 60% of seedling damage was attributed to the

senescing fronds of understorey tree ferns (Drake & Pratt

2001). In a New Zealand subtropical forest, Gillman &

Ogden (2005) reported that a large proportion of seedlings

beneath a tree fern (Cyathea dealbata) suffered physical

damage from falling fronds, whereas seedlings under

another tree fern (Dicksonia fibrosa), which retained dead

fronds, were free of litter fall damage. The physical impact

of falling senescing foliage on seedlings is negligible if the

understorey species is marcescent (Fig. 6).

Marcescence can also reduce the stress associated with

litter smothering, also a main seedling mortality agent in

forest ecosystems (Jackson & Bach 1999; Song 2007). The

stress of litter smothering is positively related to the

amount of litter falling or accumulating on the forest floor,

and leaf-shedding understorey species can contribute sub-

stantially to litter input on the forest floor (Farris-Lopez

et al. 2004; Gillman & Ogden 2005). In the present study,

with senescing fronds retained, litter input contributed by

Lastreopsis was eliminated so that seedling mortality due to

LSa was consequently reduced (Fig. 6).

The densely arranged foliage of Lastreopsis could also

contribute to reduce damage to seedlings from litter fall

and LSa through intercepting litter falling from the over-

storey canopy (Figs 3b, 4b). Many fern species are efficient

litter interceptors. Dearden & Wardle (2008) reported that

Blechnum discolor fronds, which have a funnel-like arrange-

ment, can retain up to 10% of the litter falling from the

overstorey. With the combination of short-creeping rhi-

zomes and fronds 50–100-cm long and 20–50-cm wide

(Chaffey 1999), Lastreopsis creates a continuous network of

fern stalks and overlapping fronds in the forest understo-

rey. The dense Lastreopsis canopy may displace falling litter

laterally and further protect seedlings from litter fall physi-

cal damage and LSa (Fig. 6).

Combination of species traits of fern and birds

The combination of species traits of the fern understorey

(short stature and dense foliage arrangement) and the

ground-dwelling birds (small body size and non-folivorous

diet) may contribute greatly to the positive net effect of the

understorey vegetation on uncaged seedlings (Fig. 6). The

protective effect of Lastreopsis might not be observed if

the Albert’s lyrebird and Australian brush-turkey were

folivores, as the foliage of the understorey vegetation

would be likely to attract and harbour folivores, thus

increasing the exposure of adjacent seedlings to herbivory.

These two birds are unlikely to trigger such a consumer-

mediated interaction between understorey vegetation and

tree seedlings, because they do not feed on foliage but on

invertebrates and seeds in soil and litter (Simpson & Day

1999).

The species traits of short stature and dense foliage

arrangement mean that Lastreopsis creates a physical bar-

rier and impedes the access of birds of similar stature

(Figs 3a, 4a). In a secondary eucalypt forest, Ashton &

Bassett (1997) showed that forest floor disturbance caused

by the foraging activities of the Superb lyrebird (Menura

novae-hollandiae), a species genetically and behaviourally

similar to Albert’s lyrebird, was consistently lower in areas

with dense ground fern cover (Polystichum proliferum and

Blechnumwattsii, height <1.3 m) than in bare areas. In con-

trast, in the same forest, the presence of a tree fern (Cyathea

australis, height >2 m) had no effect on lyrebird foraging

because of its relatively tall stature (Ashton & Bassett

1997). The barrier effect of understorey vegetation on the

animal species depends on their relative stature and this

effect may be maximal if their stature is similar. In the

Fig. 6. Conceptual diagram illustrating how species traits of the fern understorey (Lastreopsis decomposita) and the two ground-dwelling birds

(Menura alberti and Alectura lathami) increase seedling survival. LSa, litter smothering due to abiotic disturbance; LSb, litter smothering due to bird

disturbance.

Journal of Vegetation Science854 Doi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science

Ferns, animal non-trophic effects and seedling survival G.-Z. M. Song et al.

present study, Lastreopsis, Albert’s lyrebird and Australian

brush-turkey are all approximately 0.5–1.0 m in height

(Chaffey 1999; Simpson & Day 1999), so that Lastreopsis

could significantly impede the approach of these two bird

species (Figs 3a, 4a).

Favourable sites for tree regeneration

The most favourable areas for tree regeneration in our

study site were habitats containing both the fern understo-

rey and canopy gaps. The fern understorey may protect

seedlings from bird disturbance and litter smothering

(Figs 3a, b, 4) and in turn facilitate seedling survival

(Fig. 3c). Nevertheless, tree regeneration is likely to be

inhibited through the effective interception of light by the

fern understorey (Fig. 1). In our study site, the majority of

seedlings that developed true leaves were found in the

northernmost 10-m section of the 200-m seedling transect,

where there was a small canopy gap above (openness

10.9%) the fern understorey (Song 2007). This suggests

that extra light from canopy gaps can ease light deficits

present under the fern crown and thus promote seedling

growth.

Improved understanding of interactions between

plants and animals can promote the development of

environmentally benign and also cost-effective tools for

ecosystem management (Brooker et al. 2008). Our find-

ings indicate that, for forests harbouring organisms with

traits similar to those observed here, the facilitation of

early-stage tree regenerations might not require the use

of herbicides and the removal of ground-dwelling verte-

brates. That is, keeping rather than removing understorey

vegetation to protect seedlings from vertebrate distur-

bances and creating small canopy gaps so that seedlings

under the understorey vegetation can still have enough

light to grow should be an effective and low-impact forest

regeneration procedure.

Acknowledgements

The funding for fieldwork was from the Rainforest CRC,

Australia. Proficiency in multivariate analysis (acquired

in the Symposium and Analytic Workshop on Quantita-

tive Ecology, Fushan, Taiwan, 26 July—14 August

2007) was most helpful for the present study. The

results were initially presented at the 51st Annual Sym-

posium of the International Association for Vegetation

Science, Stellenbosch, South Africa, 7–12 September

2008. We appreciated financial support from the Oppen-

heimer Memorial Trust (Johannesburg, South Africa) to

G.-Z.M.S. for attendance at this symposium. We thank

Dr. Kuo-Jung Chao who provided valuable comments

on this manuscript.

References

Ashton, D.H. & Bassett, O.D. 1997. The effects of foraging by the

Superb Lyrebird (Menura novae-hollandiae) in Eucalyptus regn-

ans forests at Beenak, Victoria. Australian Journal of Ecology

22: 383–394.

Blundell, A.G. & Peart, D.R. 2001. Growth strategies of a shade-

tolerant tropical tree: the interactive effects of canopy gaps

and simulated herbivory. Journal of Ecology 89: 608–615.

Boogert, N.J., Paterson, D.M. & Laland, K.N. 2006. The implica-

tions of niche construction and ecosystem engineering for

conservation biology. BioScience 56: 570–578.

Bridgeland, W.T., Beier, P., Kolb, T. & Whitham, T.G. 2010. A

conditional trophic cascade: birds benefit faster growing trees

with strong links between predators and plants. Ecology 91:

73–84.

Brooker, R.W., Maestre, F.T., Callaway, R.M., Lortie, C.L., Cavi-

eres, L.A., Kunstler, G., Liancourt, P., Tielborger, K., Travis,

J.M.J., Anthelme, F., Armas, C., Coll, L., Corcket, E., Delzon,

S., Forey, E., Kikvidze, Z., Olofsson, J., Pugnaire, F., Quiroz,

C.L., Saccone, P., Schiffers, K., Seifan, M., Touzard, B. &

Michalet, R. 2008. Facilitation in plant communities: the

past, the present, and the future. Journal of Ecology 96: 18–34.

Caccia, F.D., Chaneton, E.J. & Kitzberger, T. 2009. Direct and

indirect effects of understorey bamboo shape tree regenera-

tion niches in a mixed temperate forest. Oecologia 161:

771–780.

Callaway, R.M., Kikodze, D., Chiboshvili, M. & Khetsuriani, L.

2005. Unpalatable plants protect neighbors from grazing and

increase plant community diversity. Ecology 86: 1856–1862.

Chaffey, C.H. 1999. Australian ferns: growing them successfully.

Kangaroo Press, East Roseville, NSW, AU.

Clark, D.B. & Clark, D.A. 1991. The impact of physical damage

on canopy tree regeneration in tropical rain-forest. Journal of

Ecology 79: 447–457.

Dearden, F.M. & Wardle, D.A. 2008. The potential for forest

canopy litterfall interception by a dense fern understorey,

and the consequences for litter decomposition. Oikos 117:

83–92.

Debski, I., Burslem, D.F.R.P. & Lamb, D. 2000. Ecological

processes maintaining differential tree species distributions

in an Australian subtropical rain forest: implications for

models of species coexistence. Journal of Tropical Ecology 16:

387–415.

Denslow, J.S., Newell, E. & Ellison, A.M. 1991. The effect of

understory palms and cyclanths on the growth and survival

of Inga seedlings. Biotropica 23: 225–234.

Drake, D.R. & Pratt, L.W. 2001. Seedling mortality in Hawaiian

rain forest: the role of small-scale physical disturbance.

Biotropica 33: 319–323.

Farris-Lopez, K., Denslow, J.S., Moser, B. & Passmore, H. 2004.

Influence of a common palm, Oenocarpus mapora, on seedling

establishment in a tropical moist forest in Panama. Journal of

Tropical Ecology 20: 429–438.

Journal of Vegetation ScienceDoi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science 855

G.-Z. M. Song et al. Ferns, animal non-trophic effects and seedling survival

George, L.O. & Bazzaz, F.A. 1999a. The fern understory as an

ecological filter: emergence and establishment of canopy-

tree seedlings. Ecology 80: 833–845.

George, L.O. & Bazzaz, F.A. 1999b. The fern understory as an

ecological filter: growth and survival of canopy-tree seed-

lings. Ecology 80: 846–856.

Gillman, L.N. & Ogden, J. 2005.Microsite heterogeneity in litter-

fall risk to seedlings. Austral Ecology 30: 497–504.

Gillman, L.N., Ogden, J., Wright, S.D., Stewart, K.L. & Walsh,

D.P. 2004. The influence of macro-litterfall and forest

structure on litterfall damage to seedlings. Austral Ecology

29: 305–312.

Harms, K.E., Powers, J.S. & Montgomery, R.A. 2004. Variation

in small sapling density, understory cover, and resource

availability in four Neotropical forests. Biotropica 36: 40–51.

Hegarty, E.E. 1991. Leaf litter production by lianes and trees in a

subtropical Australian rain forest. Journal of Tropical Ecology

7: 201–214.

Hurlbert, S.H. 1984. Pseudoreplication and the design of

ecological field experiments. Ecological Monographs 54:

187–211.

Ickes, K., Dewalt, S.J. & Appanah, S. 2001. Effects of native pigs

(Sus scrofa) on woody understorey vegetation in a Malaysian

lowland rain forest. Journal of Tropical Ecology 17: 191–206.

Jackson, R.V. & Bach, C.E. 1999. Effects of herbivory on

growth and survival of seedlings of a rainforest tree, Alphi-

tonia whitei (Rhamnaceae). Australian Journal of Ecology 24:

278–286.

Kikvidze, Z., Khetsuriani, L., Kikodze, D. & Callaway, R.M.

2006. Seasonal shifts in competition and facilitation in subal-

pine plant communities of the central Caucasus. Journal of

Vegetation Science 17: 77–82.

Kleinbaum, D.G. 1996. Survival analysis: a self-learning text.

Springer, New York, NY, US.

Krebs, C.J. 2001. Ecology: the experimental analysis of distribution

and abundance. Benjamin Cummings, San Francisco, CA, US.

Legendre, P. & Gallagher, E.D. 2001. Ecologically meaningful

transformations for ordination of species data. Oecologia 129:

271–280.

Levine, J.M. 1999. Indirect facilitation: evidence and predictions

from a riparian community. Ecology 80: 1762–1769.

Maestre, F.T., Valladares, F. & Reynolds, J.F. 2005. Is the change

of plant–plant interactions with abiotic stress predictable?

A meta-analysis of field results in arid environments. Journal

of Ecology 93: 748–757.

Montgomery, R.A. 2004. Effects of understory foliage on pat-

terns of light attenuation near the forest floor. Biotropica 36:

33–39.

Orrock, J.L., Holt, R.D. & Baskett, M.L. 2010. Refuge-mediated

apparent competition in plant–consumer interactions.

Ecology Letters 13: 11–20.

Osem, Y., Perevolotsky, A. & Kigel, J. 2007. Interactive effects of

grazing and shrubs on the annual plant community in semi-

arid Mediterranean shrublands. Journal of Vegetation Science

18: 869–878.

Palmer, M.W. 1988. Fractal geometry: a tool for describing spa-

tial patterns of plant communities. Vegetatio 75: 91–102.

Penfold, G.C. & Lamb, D. 1999. Species co-existence in an Aus-

tralian subtropical rain forest: evidence for compensatory

mortality. Journal of Ecology 87: 316–329.

Rebollo, S., Milchunas, D.G. & Noy-Meir, I. 2005. Refuge effects

of a cactus in grazed short-grass steppe. Journal of Vegetation

Science 16: 85–92.

Royo, A.A. & Carson, W.P. 2006. On the formation of dense

understory layers in forests worldwide: consequences and

implications for forest dynamics, biodiversity, and succes-

sion. Canadian Journal of Forest Research 36: 1345–1362.

Simpson, K. & Day, N. 1999. Field guide to the birds of Australia.

Viking, Ringwood, Vic., AU.

Song, G.Z.M. 2007. Patterns of development of regeneration of tree

species in a subtropical rainforest. PhD Thesis, The University of

Queensland, Brisbane, Qld, AU.

Spencer, R., Barley, A. & Pearson, S. 1995. Horticultural flora of

South-Eastern Australia: ferns, conifers & their allies. University

of New SouthWales Press, Sydney, NSW, AU.

Thompson, L., Thomas, C.D., Radley, J.M.A., Williamson, S. &

Lawton, J.H. 1993. The effect of earthworms and snails in a

simple plant community. Oecologia 95: 171–178.

van Veen, F.J.F., Brandon, C.E. & Godfray, H.C.J. 2009. A posi-

tive trait-mediated indirect effect involving the natural ene-

mies of competing herbivores.Oecologia 160: 195–205.

Walker, J.F., Miller, O.K., Lei, T., Semones, S., Nilsen, E. &

Clinton, B.D. 1999. Suppression of ectomycorrhizae on

canopy tree seedlings in Rhododendron maximum

L-(Ericaceae) thickets in the southern Appalachians.

Mycorrhiza 9: 49–56.

Wardle, D.A., Nilsson, M.C., Gallet, C. & Zackrisson, O. 1998. An

ecosystem-level perspective of allelopathy. Biological Reviews

73: 305–319.

Wilby, A., Shachak, M. & Boeken, B. 2001. Integration of eco-

system engineering and trophic effects of herbivores. Oikos

92: 436–444.

Wright, J.P. & Jones, C.G. 2006. The concept of organisms as

ecosystem engineers ten years on: progress, limitations, and

challenges. BioScience 56: 203–209.

Yamasaki, S.H., Fyles, J.W. & Titus, B.D. 2002. Interactions

among Kalmia angustifolia, soil characteristics, and the

growth and nutrition of black spruce seedlings in two boreal

Newfoundland plantations of contrasting fertility. Canadian

Journal of Forest Research-Revue Canadienne De Recherche

Forestiere 32: 2215–2224.

Supporting Information

Additional supporting information may be found in the

online version of this article:

Appendix S1. The effects of nine factors on survival

of uncaged and caged seedlings.

Journal of Vegetation Science856 Doi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science

Ferns, animal non-trophic effects and seedling survival G.-Z. M. Song et al.

Appendix S2. The contribution of the six mortality

agents to seedling mortality (%) under different levels of

fern cover.

Figure S1. Acquisition for light transmittance of fern

crown.

Please note: Wiley-Blackwell are not responsible for

the content or functionality of any supporting materials

supplied by the authors. Any queries (other than missing

material) should be directed to the corresponding author

for the article.

Journal of Vegetation ScienceDoi: 10.1111/j.1654-1103.2012.01398.x© 2012 International Association for Vegetation Science 857

G.-Z. M. Song et al. Ferns, animal non-trophic effects and seedling survival