Oecologia (2009) 161:771–780 DOI 10.1007/s00442-009-1412-z

C O M M U N I T Y E C O L O G Y - O RI G I N A L P A P E R

Direct and indirect eVects of understorey bamboo shape tree regeneration niches in a mixed temperate forest Fernando D. Caccia · Enrique J. Chaneton · Thomas Kitzberger

Received: 30 October 2008 / Accepted: 18 June 2009 / Published online: 10 July 2009 © Springer-Verlag 2009

Abstract Plant cover plays a major role in shaping the nature of recruitment microsites through direct (resource mediated) and indirect (consumer mediated) interactions. Understorey plants may diVerentially aVect seedling establishment, thus contributing to regeneration-niche separation among canopy tree species. We examined patterns of early tree seedling survival resulting from interactive eVects of understorey bamboo (Chusquea culeou) and resident consumers in a mixed temperate Patagonian forest, Argentina. Newly germinated seedlings of Nothofagus dombeyi and Austrocedrus chilensis were planted in bamboo thickets and non-bamboo patches, with or without small-vertebrate exclosures. We found species-speciWc patterns of seedling survival in relation to bamboo cover. Nothofagus survival was generally low but increased under bamboo, irrespective of cage treatment. Desiccation stress accounted for most Nothofagus mortality in open, non-bamboo areas. In contrast, Austrocedrus survival was highest in non-bamboo microsites, as most seedlings beneath bamboo were killed

by small vertebrates through direct consumption or nontrophic physical damage. There was little evidence for a negative impact of bamboo on tree seedling survival attributable to resource competition. The balance of simultaneous positive and negative interactions implied that bamboo presence facilitated Nothofagus early establishment but inhibited Austrocedrus recruitment via apparent competition. These results illustrate the potential for dominant understorey plants to promote microsite segregation during early stages of recruitment between tree seedlings having diVerent susceptibilities to water stress and herbivory. We recognise, however, that patterns of bamboo–seedling interactions may be conditional on moisture levels and consumer activity during establishment. Hence, both biotic and abiotic heterogeneity in understorey environments should be incorporated into conceptual models of regeneration dynamics and tree coexistence in forest communities. Keywords Apparent competition · Facilitation · Patagonia · Recruitment · Seedling predation

Communicated by Jacqui ShykoV.

Introduction F. D. Caccia (&) Departamento de Producción Vegetal, Facultad de Agronomía, Universidad de Buenos Aires, Av. San Martín 4453, 1417 Buenos Aires, Argentina e-mail: [email protected] E. J. Chaneton IFEVA-CONICET, Facultad de Agronomía, Universidad de Buenos Aires, Av. San Martín 4453, 1417 Buenos Aires, Argentina T. Kitzberger INIBIOMA-CONICET, Laboratorio Ecotono and Universidad Nacional del Comahue, Quintral 1250, 8400 Bariloche, Argentina

Seeds dispersed from parent plants land on a checkerboard of microsites created by Wne-scale heterogeneity in the abiotic and biotic environment. As a result, newly emerged seedlings face a variety of hazards, which operate as ecological Wlters to establishment (Harper 1977; George and Bazzaz 1999), and may ultimately inXuence plant community structure (Silvertown 2004). Vegetation cover plays a major role in shaping the nature of recruitment microsites, by modifying microhabitat conditions and the balance of positive (facilitation) and negative (competition, predation) interactions for emerging seedlings (McAuliVe 1986; Gill

123

772

and Marks 1991; Callaway and Walker 1997; Holmgren et al. 1997; Rousset and Lepart 2000). Thus, investigating patterns of seedling survival in relation to direct and indirect eVects from established plants may be crucial to understand diVerences in the regeneration niche of co-occurring species (Grubb 1977; Silvertown 2004). In forest communities, the presence of a dense understorey may diVerentially aVect seedling establishment of canopy tree species (Royo and Carson 2006). Yet, few studies have focused on the multiple mechanisms underlying species-speciWc eVects of understorey plants on tree seedling survival (George and Bazzaz 1999; Beckage and Clark 2003, 2005; Royo and Carson 2008). By contrast, much work has been devoted to comparing recruitment patterns between closed understorey and treefall gaps (e.g. Beckage et al. 2000; Chacón and Armesto 2006; Norghauer et al. 2008) or among diVerent gap sizes (Pearson et al. 2003). Spatial heterogeneity in understorey cover creates a forest-Xoor mosaic of recruitment opportunities through its inXuence on abiotic conditions (George and Bazzaz 1999; Beckage et al. 2000; Beckage and Clark 2003; Giordano et al. 2009), litter deposition (Gray and Spies 1997; Christie and Armesto 2003), and habitat use by resident consumers (Wada 1993; Abe et al. 2001; Pearson et al. 2003; Caccia et al. 2006; Royo and Carson 2008). Understorey vegetation may constrain tree seedling growth and survival by reducing light availability (Beckage et al. 2000; Taylor et al. 2004) and belowground resources (Lewis and Tanner 2000; Beckage and Clark 2003). Alternatively, understorey microsites with enhanced moisture or nutrient levels could facilitate seedling establishment (Callaway and Walker 1997; Heinemann and Kitzberger 2006). Critically, the outcome of direct understorey-tree seedling interactions will depend on the species’ requirements and tolerances to various stress factors (Holmgren et al. 1997). In addition, dominant understorey plants can inXuence early tree survival via consumer-mediated indirect eVects. The presence of certain plant species often increases the rate of herbivory on neighbouring plants by providing food or shelter to consumers, an indirect interaction termed “apparent competition” (Connell 1990; Chaneton and Bonsall 2000). Connell (1990) argued that apparent competition may confound the role of direct plant–plant interactions, and called for more studies manipulating consumer as well as neighbour presence in factorial designs (e.g. Gill and Marks 1991; Reader 1992; Ostfeld et al. 1997; Royo and Carson 2008). Indeed, certain plant species can induce positive, rather than negative, associational eVects by reducing herbivore incidence and damage rates on heterospeciWc neighbours (McAuliVe 1986; Agrawal et al. 2006; GraV et al. 2007). While the role of such indirect eVects in plant communities is increasingly recognised (Chaneton and Bonsall 2000; Agrawal et al. 2006), they have yet to be

123

Oecologia (2009) 161:771–780

explicitly considered as factors shaping the regeneration niches of sympatric plant species. Both vertebrate and invertebrate consumers are important agents of seedling mortality and may respond to understorey patchiness. Small mammals usually prefer vegetated patches, where they exert increased predation on seeds or seedlings relative to open areas (Wada 1993; Caccia and Ballaré 1998; Abe et al. 2001; Beckage and Clark 2005). Understorey patches also oVer resources to ground birds (Reid et al. 2004), which may inXict physical damage on young seedlings while searching for prey (Gillman and Ogden 2003). In contrast, forest insects may not exhibit a prevalent pattern of habitat use. Higher insect herbivory has been found in understorey compared to treefall gap areas (Dudt and Shure 1994), but the reversed trend has also been reported (Chacón and Armesto 2006; Richards and Cooley 2007; Norghauer et al. 2008). This reXects the inXuence of several conXicting factors on insect abundance and feeding behaviour, including light and thermal environment, plant productivity and quality, and natural enemies, which can all vary between shaded and sunlit areas (Dudt and Shure 1994; Louda and Rodman 1996; Niesenbaum and Kluger 2006; Richards and Coley 2007). Thus understorey cover may be expected to inXuence the risk of insect herbivory for emerging tree seedlings (Pearson et al. 2003). Bamboo species (Poaceae, Bambusoideae) achieve high dominance in the understorey of many temperate forests (Veblen et al. 1996; Takahashi 1997; Abe et al. 2001; Taylor et al. 2004). Understorey bamboos have been reported to inhibit tree regeneration through light competition (Veblen et al. 1996, Takahashi 1997), low red:far-red ratios (Giordano et al. 2009), and increased seed predation (Wada 1993) or fungal attack on seedlings (Abe et al. 2001). Yet, how diVerent processes acting in concert aVect tree recruitment in bamboo understories is largely unknown. Previous work in native cool-temperate forests of the Patagonian Andes in southern South America showed that bamboo cover (Chusquea spp.) increases tree seed losses to granivores, and that predation is much higher for the large-seeded Austrocedrus chilensis than for the small-seeded Nothofagus dombeyi (Caccia et al. 2006). Small mammals are major seed predators in these forests (Kitzberger et al. 2007), but little is known about their impact on seedlings (Caccia and Ballaré 1998). Here, we focus on the establishment phase of the regeneration niche by addressing the interactive eVects of bamboo and resident consumers on early tree seedling survival in a mixed Patagonian forest. We conducted a Weld transplant experiment to determine the relative strength and net balance of direct (resource mediated) and indirect (consumer mediated) eVects of bamboo on newly emerged seedlings of two canopy trees, N. dombeyi and A. chilensis. Species-speciWc responses to variation in biotic/abiotic environment between bamboo

Oecologia (2009) 161:771–780

and non-bamboo microsites may lead to early segregation of regeneration niches (Grubb 1977; Takahashi 1997). SpeciWcally, we asked the following questions: 1. How does bamboo cover alter microhabitat conditions and tree seedling survival irrespective of consumer pressure? 2. Do consumers mediate the eVect of bamboo on early seedling establishment? 3. How does the sign and magnitude of consumer-mediated indirect eVects change with seedling species and herbivore type (vertebrate or invertebrate)? 4. What is the net balance of direct and indirect bamboo eVects for spatial patterns of early seedling survival in diVerent tree species?

Materials and methods Study system The study was conducted in a mixed N. dombeyi–A. chilensis forest stand located at 870 m a.s.l. on an east-facing slope overlooking the Casa de Piedra creek, about 10 km west of Bariloche, Nahuel-Huapi National Park, Argentina (41°07⬘59⬙S, 71°27⬘11⬙W). The site is representative of low-elevation, evergreen, sub-humid mixed forests extending along the eastern foothills of the Andes in northwestern Patagonia (Veblen et al. 1996). The climate is characterised by cold and wet winters, and mild but dry summers. Mean annual precipitation is »1,600 mm, mostly occurring as rain and snow during autumn and winter (March–September). Mean monthly temperatures vary from 3.2°C in July to 14°C in January (1984–2003, INTA weather station, ca.15 km from study site). Soils are poorly developed Andisols derived from volcanic ashes. N. dombeyi is the dominant tree species (>70%). Understorey vegetation is dominated by the native bamboo Chusquea culeou, which forms dense thickets interspersed with open patches lacking bamboo and having a sparse herbaceous cover. Bamboo patches may reach densities up to 30–60 culms m¡2 and may grow up to 5 m high (Pearson et al. 1994). Several shrub species (mainly Schinus patagonicus, Aristotelia chilensis and Berberis darwinii) were scattered in low numbers through the forest understorey. The study area lacked Wre for at least a century; no signs of cattle activity were found during the course of the experiment. The main growing season extends from October to April. Synchronous seedfall for canopy tree species occurs during autumn, from late March to late April; tree seedling emergence usually takes place in late spring, between November and early December. Newly emerged seedlings may be killed by vertebrate or invertebrate

773

consumers. Potential seedling predators include the omnivorous rodents Abrothrix olivaceus, Abrothrix longipilis and Geoxus valdivianus (Murúa 1996; Silva 2005). Insect activity increases over the growing season, with maximum damage levels from canopy folivores occurring in late summer (Mazía et al. 2004). At the forest Xoor, tree seedlings may be consumed by various insect herbivores (Marchesini 2006). Lastly, seedlings may be also killed by non-trophic animal damage associated with foraging activities of ground birds (Burschel et al. 1976; Reid et al. 2004) and small mammals (Gillman and Ogden 2003). The study was conducted between 25 November 2003 (mid spring) and 20 April 2004 (mid autumn), before the Wrst snowfall. Although climatic conditions at the start of the experiment in November and December 2003 were relatively wet (total rainfall = 76.1 mm, 48% above average), the ensuing summer (January–March) was very dry (21 mm, 70% below average), with no precipitation during January and February, and was also warmer than usual (15.7°C, mean = 13.3°C). Experimental design We performed a Weld transplant experiment using newly germinated seedlings of N. dombeyi and A. chilensis (hereafter named only by genus). The experiment was set up as a three-way factorial design including two understorey habitats (bamboo present vs. absent), two predation levels (cage exclosure vs. uncaged control), and two seedling densities (two vs. ten seedlings). Although prior work showed that granivory rates on these tree species were largely density independent (Caccia et al. 2006), we decided to include density as a treatment because of its potential to alter bamboo–seedling interactions through consumer-mediated eVects (Janzen 1970; Clark and Clark 1985). Each experimental unit consisted of two or ten seedlings placed under dense bamboo cover or in non-bamboo areas, with or without an exclusion cage. These densities were intended to simulate natural variation in seedling emergence (Blackhall et al. 2008). Tree species were always transplanted into separate microsites. Exclusion cages were made of 1-cmmesh galvanised wire (15 cm £ 15 cm £ 15 cm) and were pushed »8 cm into the soil. Cages prevented seedlings from being consumed or damaged by vertebrates (rodents or birds) but did not exclude most invertebrate herbivores. Open understorey patches were 5–8 m in diameter and had no shrubs. Seedling patches were located at least 5 m apart. The minimum distance between seedling patches located in adjacent bamboo and non-bamboo patches was »5 m. Thus, for each species, the experiment comprised eight treatments arranged in a fully randomised design, with 10 and 20 replicates for the caged and uncaged treatments,

123

774

respectively. This yielded a total of 240 experimental units (120 units per species). Replication was doubled for the uncaged treatments because consumer impact on seedling survival may be highly variable among understorey areas (Caccia and Ballaré 1998). Austrocedrus seeds were germinated in 250-cm3 pots after 60 days of cold–wet stratiWcation. Before being transferred to the Weld, newly germinated seedlings were grown in an unheated greenhouse and watered daily from August to November 2003 until they attained »2 cm height and had developed the Wrst set of needle-shaped primary leaves. Because Nothofagus seeds were not available in suYcient numbers, we used freshly emerged seedlings which were all collected in November 2003 from nearby forests (within 3 km of the study site). Nothofagus seedlings »1 cm height and with two fully expanded cotyledons were carefully removed and planted in trays Wlled with local soil. These seedlings were kept in the greenhouse for 10 days before starting the experiment. Only vigorous seedlings with no apparent damage were used. On 25 November 2003, tree seedlings were transplanted to the Weld with a »7-cm-deep soil core into holes made in the soil, leaving 2–4 cm between seedlings. The litter layer was replaced around the seedling patches after planting. Each experimental unit received »250 ml of water to reduce transplant shock; no additional watering was supplied during the course of the experiment. Only three (2.5%) Nothofagus seedlings died within 10 days of planting and were immediately replaced, as it was assumed that they were killed during transplanting. Seedlings were censused 4 times, on 5 December 2003 (10 days after planting) and on 16 January, 20 March and 20 April 2004. We counted the number of plants surviving per microsite and determined the apparent causes of seedling death (see below). In addition, we recorded the amount of leaf area damaged by folivorous insects (Mazía et al. 2004; Marchesini 2006). Damage was assessed by scoring the percent area missing or damaged for each leaf or cotyledon using a Wve-point scale. These measurements were applied only to Nothofagus, since very few Austrocedrus seedlings were found with damaged leaves. Seedling mortality was attributed to abiotic stress (desiccation) or predation. We further attempted to distinguish between seedlings killed by rodents or insects. Overall, seedlings killed by herbivores comprised four categories: seedlings with a clean cut at the hypocotyl level (Ct), uprooted seedlings (Up), seedlings with all leaf area removed and only the stem left (Sl), and missing (Mi). While seedlings in categories Ct, Sl and Mi occurred both inside and outside the cage exclosures, Up seedlings were found only outside the cages, indicating they were most likely damaged by small vertebrates. To estimate mortality by invertebrate consumers, we considered only seedlings

123

Oecologia (2009) 161:771–780

killed inside the cages. Seedling mortality by small vertebrates was estimated as the diVerence between the number of plants killed outside and inside cages (Ct + Sl + Mi) plus any uprooted seedlings. Because the design was fully randomised (cage and control microsites were not paired), cumulative mortality for each uncaged microsite (n = 20) was compared against the average mortality in the caged microsites (attributed to invertebrates) for each bamboodensity treatment. Microenvironmental measurements To characterise microsite conditions for tree seedlings, in February 2004, we measured light intensity, soil moisture, air and soil temperatures, and litter depth in randomly selected bamboo and non-bamboo understorey patches. Measurements were taken at 1–2 h intervals, between 1100 and 1600 hours, for two consecutive sunny days. Photosynthetic photon Xux density (PPFD) was recorded at 5 cm height above the soil level in 27 bamboo and 33 non-bamboo patches (3 readings/patch) using a quantum sensor attached to a LI-1000 radiometer (Li-Cor, Neb.). Volumetric soil moisture content (%) was measured to 7-cm soil depth in 29 bamboo and 32 non-bamboo patches (4 readings/patch)using a Theta Probe sensor (Delta-T Devices, Cambridge, UK). The depth of the surface litter layer was measured to the nearest 0.1 cm (n = 30 patches; 3 readings/ patch), together with the soil moisture measurements. In addition, we recorded daytime air, litter, and soil temperatures on an hourly basis in two to four patches of each cover type (n = 32 for soil and litter, and n = 36 for air temperatures; 2 readings/patch) using a digital thermometer. Temperature readings were taken 10 cm above soil level (air temperature), within the litter layer, and at 2-cm depth in the mineral soil. Statistical analyses Seedling survival data were analysed separately for each tree species using generalised linear modelling (PROC GENMOD; SAS Institute 1996) using a binomial error structure and a logit-link function. The total number of seedlings per microsite was used as binomial denominator. Treatment eVects were tested through analysis of deviance (Crawley 1993). Initial models included all factor interactions and were simpliWed until all remaining parameters were signiWcant at P = 0.05. Models were checked for overdispersion by looking at the ratio between the residual scaled deviance and the residual df (Crawley 1993). Where necessary, models were adjusted using the ‘dscale’ procedure (SAS Institute 1996). We Wrst tested for eVects of bamboo cover (two levels), cage exclosure (two levels) and density (two levels) on seedling survival, or the proportion

Microenvironmental conditions The presence of bamboo signiWcantly modiWed microsite conditions at the forest Xoor level. Bamboo cover reduced average light levels to 3.5% of incoming radiation in nonbamboo understorey patches (F1,58 = 315.3, P < 0.0001; Fig. 1a). Mean temperatures for the top mineral soil (F1,62 = 199.7, P < 0.0001) and litter layer (F1,62 = 36.5, P < 0.0001) were »2°C cooler in bamboo thickets than in non-bamboo microsites (Fig. 1c). Air temperatures were also reduced by 1.5°C within bamboo canopies (F1,70 = 6.3, P = 0.014). Soil moisture content (F1,59 = 27, P < 0.0001) and litter depth (F1,58 = 32.7, P < 0.0001) were both signiWcantly higher in bamboo thickets than in open understorey areas (Fig. 1b, d). Overall pattern of seedling survival Tree species exhibited opposing patterns of seedling survival in response to bamboo cover (Fig. 2). Nothofagus survivorship was low (<20%) on average but was signiWcantly increased by bamboo cover (F1,118 = 42.4, P < 0.0001). Small-vertebrate exclosures did not aVect Nothofagus survival (cage P = 0.36, cage £ bamboo, P = 0.65). In contrast, survival of Austrocedrus seedlings varied widely among forest microsites (8–55%) and was

(a)

20

Soil moisture (%)

150 120 90 60 30 0

(b)

15 10 5 0

(c)

16

Litter depth (cm)

20 15 10 5

(d)

12 8 4 0

0 –Bamboo

+ Bamboo

–Bamboo

+ Bamboo

Fig. 1 Microenvironmental diVerences between bamboo (+Bamboo) and non-bamboo (¡Bamboo) understorey patches in a mixed Nothofagus–Austrocedrus forest. a Photosynthetic photon Xux density (PPFD; 5 cm above ground level), b volumetric soil moisture content (%), c mean hourly temperatures for the topsoil (black bars) and surface litter layer (white bars), and d plant litter depth. Data show means § 1 SE (February 2004, mid summer, n = 27–36) 60

Control Cage

50

Survivorship(%)

Results

Temperature (°C)

of seedlings surviving at the end of the experiment. In a second set of analyses, we evaluated how apparent causes of seedling death (predation or desiccation) interacted with bamboo cover, seedling density and cage treatment in determining mortality patterns. Lastly, we examined the eVect of bamboo cover and seedling density on the estimated proportions of seedlings killed by vertebrate and invertebrate consumers. For Nothofagus seedlings, patterns of leaf damage by insects were examined using three-way ANOVA (PROC GLM; SAS Institute 1996) with bamboo cover, cage exclosure and seedling density as main eVects. The mean leaf area damaged (%) per plant surviving at the end of the experiment was used as response variable. Damage data were arcsine square-root transformed to reduce variance heterogeneity and meet the normality assumption. We used one-way ANOVA to test for diVerences in microenvironmental variables between bamboo and non-bamboo patches. Analyses were performed on the mean values obtained for each replicate patch after averaging across all readings taken at 1–2 h intervals over two consecutive days. Light (PPFD) and soil temperature data were logtransformed before analyses.

775

PPFD (µmol m-2 s -1)

Oecologia (2009) 161:771–780

40 Austrocedrus 30 20 10

Nothofagus

0 –Bamboo

+ Bamboo

Fig. 2 EVects of bamboo cover and vertebrate exclusion on Nothofagus (dashed line) and Austrocedrus (solid line) seedling survival. Values are means of 10 (caged) or 20 (control) microsites, pooled over two seedling-density levels. There were signiWcant eVects of bamboo for Nothofagus (P < 0.0001), and bamboo £ cage interaction for Austrocedrus (P < 0.002)

highest in non-bamboo patches (F1,116 = 24.15, P < 0.0001). Vertebrate exclosures positively inXuenced Austrocedrus survivorship (F1,116 = 16.5, P < 0.0001), although the magnitude of the eVect depended strongly on bamboo cover (cage £ bamboo F1,116 = 10.47, P = 0.0016). Whereas cages had little impact on survival in non-bamboo areas, Austrocedrus seedlings had a very low chance of surviving in bamboo patches when fully exposed to predators (Fig. 2). For both species, seedling survival was independent of initial density (F1,115 = 0.97, P = 0.33 and F1,115 = 0.51, P = 0.48 for Nothofagus and Austrocedrus, respectively; for both species, all treatment interactions with density had P-values > 0.10).

123

776

Oecologia (2009) 161:771–780

Table 1 EVects of bamboo cover, small-vertebrate exclusion, seedling density, and source of mortality (desiccation or predation) on tree seedling mortality rates in mixed Nothofagus–Austrocedrus forest in northern Patagonia df

Nothofagus

Austrocedrus

60

F

F

40

P

P

1

0.36

0.55

8.32

0.0043

Cage exclosure (C)

1

0.01

0.91

3.17

0.077

Density (D)

1

0.22

0.64

0.25

0.62

Mortality source (M)

1

389.68

<0.0001

0.08

0.78

B£C

1

0.01

0.93

0.74

0.39

B£M

1

139.02

<0.0001

53.86

<0.0001

B£D

1

0.23

0.63

0.13

0.72

C£M

1

0.21

0.65

8.36

0.0042

C£D

1

0.09

0.76

<0.01

0.99

20

M£D

1

<0.01

0.97

0.09

0.76

0

B£C£M

1

2.02

0.16

6.04

0.0147

Other three-way interactions

3

<0.63

>0.43

<2.15

>0.15

Four-way interaction

1

0.04

0.84

<0.01

0.99

Parameters with a signiWcant (P < 0.05) contribution to a Wnal model are shown in bold (residual df = 236 and 232, for Nothofagus and Austrocedrus, respectively). EVects were tested through analysis of deviance assuming binomial errors

Austrocedrus

(b)

Control Cage

20

Bamboo cover (B)

a

(a)

80

Seedling mortality (%)

Parameter

a

Nothofagus 100

0 100

(c)

(d)

80 60 40

–Bamboo

+ Bamboo

–Bamboo

+ Bamboo

Fig. 3 Tree seedling mortality by desiccation (a, b) and predation (c, d) as inXuenced by bamboo cover and vertebrate exclusion. c, d Include seedlings killed by either vertebrate or invertebrate consumers. Values are means of 10 (caged) or 20 (control) microsites, pooled over two seedling-density treatments. There were signiWcant eVects of bamboo £ mortality source (P < 0.0001) for Nothofagus, and bamboo £ cage £ mortality source (P < 0.02) for Austrocedrus

Sources of tree seedling mortality The relative role of biotic and abiotic sources of seedling mortality for both species signiWcantly changed between bamboo and non-bamboo microsites (bamboo £ mortality source P < 0.0001; Table 1). In general, abiotic stress appeared to be the main cause of mortality in non-bamboo areas, whereas predation was a greater mortality source within bamboo thickets (Fig. 3). More importantly, the role of abiotic stress and predation in tree seedling mortality was clearly species speciWc. Desiccation stress accounted for most Nothofagus mortality, especially in non-bamboo areas, where it killed »90% of the seedlings (Table 1; Fig. 3). Total predation on Nothofagus seedlings increased in bamboo patches but did not vary with vertebrate exclusion (Table 1; Fig. 3c). For Austrocedrus, desiccation and predation were also the primary causes of seedling death in non-bamboo and bamboo microsites, respectively (see Fig. 3b, d). However, Austrocedrus seedling loss to predation in bamboo thickets was markedly reduced by vertebrate exclusion (Table 1; Fig. 3d). The cage treatment did not alter seedling mortality attributed to desiccation stress (Fig. 3a, b). Seedling density did not signiWcantly modify patterns in the source of mortality for either tree species (Table 1). We found no seedlings with signs of being killed by damping-oV.

123

Seedling mortality and predator identity Irrespective of tree species, seedling mortalities attributed to either invertebrate (Nothofagus F1,18 = 7.1, P = 0.016; Austrocedrus F1,18 = 12.2, P = 0.0026) or vertebrate consumers (Nothofagus F1,38 = 9.3, P = 0.004; Austrocedrus F1,38 = 26.2, P < 0.0001) were signiWcantly higher in bamboo thickets than in non-bamboo areas (Fig. 4). Neither seedling density nor its interaction with bamboo cover signiWcantly aVected mortality patterns associated with either invertebrate or vertebrate consumers (all P-values > 0.10 for both Nothofagus and Austrocedrus). A greater proportion of Nothofagus seedlings were killed by invertebrate herbivores (= lost from inside cages; Fig. 4a), whereas Austrocedrus seedlings were mostly killed by small vertebrates (Fig. 4b). Moreover, a large fraction of Austrocedrus deaths attributed to vertebrate predators were uprooted seedlings, and these were more common beneath bamboo than in nonbamboo areas (44 vs. 8%, F1,78 = 34.1, P < 0.0001). After 5 months of experiment, leaf area damaged by chewing insects on surviving Nothofagus seedlings averaged 1.5%. Insect-damaged seedlings were only found in bamboo patches, but treatments did not diVer signiWcantly in mean folivory rates (P > 0.10).

Oecologia (2009) 161:771–780

777

(a) Nothofagus 60

Invertebrates

Seedlings killedby herbivores (%)

Vertebrates 40

20 0

(b) Austrocedrus 60

40

20 0

– Bamboo

+ Bamboo

Fig. 4 The proportion of a Nothofagus and b Austrocedrus mortality attributed to diVerent consumers in forest understorey with or without bamboo (n = 10 and 20, for invertebrate and vertebrate predation, respectively). Data averaged over the two seedling-density treatments

Discussion Despite the potential for understorey vegetation to act as a major Wlter to recruitment (George and Bazzaz 1999; Royo and Carson 2006), few experiments have tested its multiple eVects on tree seedling establishment through resource and consumer-mediated interactions (Royo and Carson 2008). We found that bamboo presence determined spatial patterns of seedling survival through the Wrst post-emergence summer in a closed-canopy Patagonian forest. Bamboo inXuenced survivorship via a combination of direct and indirect eVects, which diVerentially impinged on two canopy species. The balance of positive and negative bamboo eVects implied that Nothofagus early survival was facilitated by bamboo. Conversely, Austrocedrus had greater chances of surviving in non-bamboo areas, due to increased seedling predation under bamboo thickets. We show that understorey cover modulated the relative strength of facilitation vs. apparent competition, and thus could potentially promote an early segregation of regeneration niches between these canopy tree species. The presence of bamboo enhanced seedling survivorship by ameliorating abiotic conditions for both tree species. More seedlings died from desiccation away from than within bamboo thickets, irrespective of predator exclusion (Fig. 3a, b). The magnitude of this facilitative eVect diVered between species, being much stronger for Nothofagus than for Austrocedrus. The latter is broadly tolerant to water

stress and dominates large expanses of drier, open woodlands in the eastern forest-steppe ecotone of the Andean forests (Veblen et al. 1996). In these drier woodlands, Austrocedrus recruitment is facilitated by shrubs during average rainfall years (Kitzberger et al. 2000). In this study, abiotic stress appeared to be the main cause of Nothofagus mortality. Thus, microhabitat amelioration by bamboo prevailed over apparent competition in determining recruitment opportunities for Nothofagus (Fig. 2). Bamboo cover reduced light penetration to the forest Xoor, creating cooler and wetter microsites, which might have protected seedlings from drought stress during mid summer. Similarly, survival of Nothofagus pumilio seedlings in high-elevation Andean forests was higher within dense forb patches than in open understorey areas (Heinemann and Kitzberger 2006). In contrast, several studies have reported negative direct eVects of understorey layers (including various bamboo species) on tree seedling survival and growth (e.g. Veblen 1982; Beckage and Clark 2003; Taylor et al. 2004; Royo and Carson 2006). Early facilitation of survival may give way to competition from understorey vegetation in later stages of tree recruitment (see Callaway and Walker 1997). Bamboo also exerted negative, consumer-mediated eVects on seedling survival. Cumulative mortality through both herbivory and non-trophic animal damage strongly altered Austrocedrus seedlings, having a comparatively small impact on Nothofagus (Fig. 3c, d). As a result, Austrocedrus survival was largely restricted to non-bamboo microsites (Fig. 2). These Wndings indicate that apparent competition from bamboo (Connell 1990) played a signiWcant, but species-speciWc role, in determining the relative safety of establishment microsites in the forest understorey (see also Royo and Carson 2008). The absence of densitydependent seedling mortality observed in this study suggests that apparent competition induced by understorey cover might operate irrespective of natural variation in seedling densities. Likewise, in this system, the intensity of granivore-mediated apparent competition associated with bamboo presence was found to be independent of seedpatch density for both Austrocedrus and Nothofagus (Caccia et al. 2006). The type of consumer mediating the indirect eVect of bamboo appeared to shift according with species identity (Fig. 4). We found a major enhancement of Austrocedrus survival when vertebrates were prevented access to seedlings in bamboo patches (Fig. 2), suggesting that small mammals were probably the primary agents of Austrocedrus mortality beneath bamboo. This is consistent with the preferential use of bamboo patches by rodents (Caccia et al. 2006), as well as by ground-foraging birds (Burschel et al. 1976; Reid et al. 2004). Small-vertebrate consumers may be attracted to bamboo thickets as they seek refuge from

123

778

predators, or because these patches oVer a better supply of alternative food items (Reid et al. 2004). Interestingly, a large proportion of Austrocedrus seedlings in bamboo patches were pulled oV (uprooted) but left uneaten, indicating that not only herbivory but also physical damage was an important component of apparent competition. Non-trophic interactions with consumers have been found to alter seedling recruitment in other plant communities (Thompson et al. 1993; Gillman and Ogden 2003). Invertebrate predation represented an additional source of seedling mortality that further limited establishment in bamboo patches. Our study is one of the few showing how invertebrate herbivory changes with understorey cover in a closed-canopy forest environment (cf. Pearson et al. 2003). The indirect impact of bamboo on seedlings protected from vertebrates (Fig. 3c, d) might have been elicited by various mechanisms. For instance, the litter layer beneath bamboo may harbour higher densities of soil-dwelling insects or molluscs, which could feed on the nutrient-rich cotyledons of young tree seedlings (Facelli 1994). Alternatively, invertebrate herbivory might be similar across understorey patches, although seedlings in bamboo microsites would be more limited by light availability (see Fig. 1) and less able to recover from damage than seedlings emerging in sunlit microsites (Clark and Clark 1985; Norghauer et al. 2008). While invertebrates killed about 30% of all tree seedlings beneath bamboo, they exerted a greater relative impact on Nothofagus, whereas vertebrates killed more Austrocedrus seedlings (Fig. 4). This was consistent with the higher folivory levels recorded on Nothofagus seedlings. Nothofagus is the dominant canopy tree in these forests, and as a broadleaf species is attacked by several insect folivores, while the needle-shaped Austrocedrus leaves may be less attractive to herbivorous insects. Overall, our results suggest that apparent competition induced by habitat-forming plants (Connell 1990; Caccia et al. 2006) may alter recruitment of co-occurring plant species via interaction chains involving diVerent herbivore guilds. Plant regeneration niches involve both abiotic and biotic environmental dimensions (Grubb 1977; Schupp 1995; Silvertown 2004). Many forests support dense understorey layers that alter conditions for seedling establishment (George and Bazzaz 1999; Royo and Carson 2006). Several studies reported only weak evidence for microhabitat specialisation among temperate tree species in relation to understorey structure (Gray and Spies 1997; Beckage and Clark 2003). We found that bamboo inXuenced the spatial pattern of early tree seedling survival through both positive and negative eVects. Moreover, the net balance of these interactions depended on the seedling species, with facilitation driving Nothofagus survival and apparent competition controlling Austrocedrus. A major consequence of these contrasting responses was the partitioning of understorey

123

Oecologia (2009) 161:771–780

microsites for potential tree regeneration. Similar neighbourhood interactions appear to contribute to the successful recruitment of co-occurring species in arid plant communities (Callaway 1992; GraV et al. 2007). Our results may contribute to understanding the early dynamics of tree regeneration in this mixed forest community. Most Austrocedrus mortality was attributable to consumers and was highest in bamboo thickets, precisely where rodents also produce maximal seed losses for this species (Caccia et al. 2006). Small mammals may thus exert a signiWcant control on the overall success and spatial distribution of Austrocedrus recruitment in this forest. By multiplying seed survival by seedling survival rates for bamboo (0.11 £ 0.07 = 0.0077) and non-bamboo (0.65 £ 0.43 = 0.28) areas, we estimate that (other things being equal) Austrocedrus would have a two orders of magnitude higher probability of recruiting outside bamboo thickets, where seedlings encounter ‘consumer-free space’ (Holt 1984). Given that seed predation in non-bamboo patches rises markedly during years of high rodent activity (Caccia et al. 2006), windows of opportunity for Austrocedrus establishment would open in understorey gaps mostly in years or sites with high tree seed crops and/or low rodent densities. On the other hand, Nothofagus may present a case of seed-seedling conXict with regard to early survival probabilities in diVerent microsites (Schupp 1995). We have shown that Nothofagus seed removal is higher under bamboo than in non-bamboo areas (Caccia et al. 2006), suggesting that the safer microsites for seeds may be the poorer ones for new seedlings and vice versa (Schupp 1995). Because in this study seedling losses for Nothofagus were much higher than seed predation rates (Caccia et al. 2006), there might be an estimated tenfold greater chance of seedlings establishing in bamboo (0.85 £ 0.11 = 0.094) than in non-bamboo areas (1.0 £ 0.01 = 0.01). If seedlings of the shade-intolerant Nothofagus manage to survive beneath bamboo, they might be capable of responding to increased light levels after tree falls (Veblen 1989; Veblen et al. 1996) or episodic die-oV of bamboo (Holz and Veblen 2006; RaVaele et al. 2007). Thus, in dry years, facilitation by bamboo could open a narrow temporal window of opportunity for Nothofagus regeneration. It is worth noting, however, that our study was performed in a subhumid forest site and during a very dry summer. Evidence shows that the outcome of plant–plant interactions, and in particular the role of facilitation in recruitment, may be conditional on interannual climatic Xuctuation (Kitzberger et al. 2000). In average (less dry) or wetter years, competition from bamboo could become the prevalent force (see Veblen et al. 1996) determining higher seedling survival in non-bamboo areas. Indeed, dense bamboo has been reported to be a major factor inhibiting Nothofagus species regeneration in

Oecologia (2009) 161:771–780

more humid Andean forests (Veblen 1989; Veblen et al. 1996; González et al. 2002). In conclusion, our study emphasises that seedling survival patterns in forest communities are not only the result of resource-mediated interactions, but may also reXect diVerential predation risk across understorey microsites (Janzen 1970; Beckage and Clark 2005; Royo and Carson 2008). Moreover, results illustrate how microsite heterogeneity created by a dominant understorey plant may promote early segregation of potential regeneration niches for canopy tree species through an array of direct and indirect eVects. We nevertheless recognise that the outcome of understoreytree seedling interactions will depend on abiotic stress levels and consumer activity during recruitment. Therefore, both biotic and abiotic variation in understorey environments should be explicitly incorporated into conceptual models of tree regeneration and coexistence in forest communities. Acknowledgments We thank the Administración de Parques Nacionales for supporting our research at Nahuel-Huapi National Park. Three reviewers provided several constructive suggestions that helped to improve the manuscript. The study was funded by grants from the Universidad de Buenos Aires (G-407) and the British Ecological Society (SEPG Nº 2029). The reported experiment complied with Argentinean laws.

References Abe M, Miguchi H, Nakashizuka T (2001) An interactive eVect of simultaneous death of dwarf bamboo, canopy gap, and predatory rodents on beech regeneration. Oecologia 127:281–286 Agrawal AA, Lau JA, Hambäck PA (2006) Community heterogeneity and the evolution of interactions between plants and insect herbivores. Quart Rev Biol 81:349–376 Beckage B, Clark JS (2003) Seedling survival and growth of three forest tree species: the role of spatial heterogeneity. Ecology 84:1849–1861 Beckage B, Clark JS (2005) Does predation contribute to tree diversity? Oecologia 143:458–469 Beckage B, Clark JS, Clinton BD, Haines BL (2000) A long-term study of tree seedling recruitment in southern Appalachian forests: the eVects of canopy gaps and shrub understories. Can J For Res 30:1617–1631 Blackhall M, RaVaele E, Veblen TT (2008) Cattle aVect early post-Wre regeneration in a Nothofagus dombeyi–Austrocedrus chilensis mixed forest in northern Patagonia, Argentina. Biol Conserv 141:2251–2261 Burschel P, Gallegos C, Martínez O, Moll W (1976) Composición y dinámica regenerativa de un bosque virgen mixto de raulí y coigue. Bosque 2:55–74 Caccia FD, Ballaré CL (1998) EVects of tree cover, understory vegetation, and litter on regeneration of Douglas-Wr (Pseudotsuga menziesii) in southwestern Argentina. Can J For Res 28:683–692 Caccia FD, Chaneton EJ, Kitzberger T (2006) Trophic and non-trophic pathways mediate apparent competition through post-dispersal seed predation in a Patagonian mixed forest. Oikos 113:469–480 Callaway RM (1992) EVect of shrubs on recruitment of Quercus douglasii and Quercus lobata in California. Ecology 73:2118– 2128

779 Callaway RM, Walker LR (1997) Competition and facilitation: synthetic approach to interactions in plant communities. Ecology 78:1958–1965 Chacón P, Armesto JJ (2006) Do carbon-based defences reduce foliar damage? Habitat-related eVects on tree seedling performance in a temperate rainforest of Chiloé Island, Chile. Oecologia 146:555– 565 Chaneton EJ, Bonsall MB (2000) Enemy-mediated apparent competition: empirical patterns and the evidence. Oikos 88:380–394 Christie DA, Armesto JJ (2003) Regeneration microsites and tree species coexistence in temperate rain forests of Chiloé Island, Chile. J Ecol 91:776–784 Clark DB, Clark DA (1985) Seedling dynamics of a tropical tree: impacts of herbivory and meristem damage. Ecology 66:1884–1892 Connell JH (1990) Apparent versus “real” competition in plants. In: Grace JB, Tilman D (eds) Perspectives on plant competition. Academic Press, San Diego, pp 9–26 Crawley MJ (1993) GLIM for ecologists. Blackwell, Oxford Dudt JF, Shure DJ (1994) The inXuence of light and nutrients on foliar phenolics and insect herbivory. Ecology 75:86–95 Facelli JM (1994) Multiple indirect eVects of plant litter aVect the establishment of woody seedlings in old Welds. Ecology 71:1727– 1735 George LO, Bazzaz FA (1999) The fern understory as an ecological Wlter: emergence and establishment of canopy-tree seedlings. Ecology 80:833–845 Gill DS, Marks PL (1991) Tree and shrub seedling colonization of old Welds in central New York. Ecol Monogr 61:183–205 Gillman LN, Ogden J (2003) Seedling mortality and damage due to non-trophic animal interactions in a northern New Zealand forest. Aust Ecol 28:48–52 Giordano CV, Sánchez RA, Austin AT (2009) Gregarious bamboo Xowering opens a window of opportunity for regeneration in a temperate forest of Patagonia. New Phytol 181:880–889 González ME, Veblen TT, Donoso C, Valeria L (2002) Tree regeneration responses in a lowland Nothofagus-dominated forest after bamboo dieback in south-central Chile. Plant Ecol 161:59–73 GraV P, Aguiar MR, Chaneton EJ (2007) Shifts in positive and negative plant interactions along a grazing intensity gradient. Ecology 88:188–199 Gray AN, Spies TA (1997) Microsite controls on tree seedling establishment in conifer forest canopy gaps. Ecology 78:2458–2473 Grubb PJ (1977) The maintenance of species richness in plant communities: the importance of the regeneration niche. Biol Rev 52:107– 145 Harper JL (1977) Population biology of plants. Academic Press, London Heinemann K, Kitzberger T (2006) EVects of position, understorey vegetation and coarse woody debris on tree regeneration in two environmentally contrasting forests of north-western Patagonia: a manipulative approach. J Biogeogr 33:1357–1367 Holmgren M, ScheVer M, Huston MA (1997) The interplay of facilitation and competition in plant communities. Ecology 78:1966– 1975 Holt RD (1984) Spatial heterogeneity, indirect interactions, and the coexistence of prey species. Am Nat 124:377–406 Holz CA, Veblen TT (2006) Tree regeneration response to Chusquea montana bamboo die-oV in a subalpine Nothofagus forest in the southern Andes. J Veg Sci 17:19–28 Janzen D (1970) Herbivores and the number of tree species in tropical forests. Am Nat 104:501–528 Kitzberger T, Steinaker DF, Veblen TT (2000) EVects of climatic variability on facilitation of tree establishment in northern Patagonia. Ecology 81:1914–1924 Kitzberger T, Chaneton EJ, Caccia FD (2007) Indirect eVects of prey swamping: diVerential seed predation during a bamboo masting event. Ecology 88:2541–2554

123

780 Lewis SL, Tanner EVJ (2000) EVects of aboveground and belowground competition on growth and survival of rain forest tree seedlings. Ecology 81:2525–2538 Louda SM, Rodman JE (1996) Insect herbivory as a major factor in the shade distribution of a native crucifer (Cardamine cordifolia A. Gray, bittercress). J Ecol 84:229–237 Marchesini VA (2006) Consecuencias ecológicas de la Xoración masiva del bambú Chusquea culeou (Poacea: Bambusoideae) en un bosque templado del sur Argentino. Master’s thesis. Faculty of Agronomy, University of Buenos Aires, Buenos Aires Mazía CN, Kitzberger T, Chaneton EJ (2004) Interannual changes in folivory and bird insectivory along a natural productivity gradient in northern Patagonian forests. Ecography 27:29–40 McAuliVe JR (1986) Herbivore-limited establishment of a Sonoran desert tree: Cercidium microphyllum. Ecology 67:276–280 Murúa R (1996) Comunidades de mamíferos del bosque templado de Chile. In: Armesto JJ, Villagrán C, Arroyo MTK (eds) Ecología de los bosques nativos de Chile. Editorial Universitaria, Santiago, pp 113–134 Niesenbaum RA, Kluger EC (2006) When studying the eVects of light on herbivory, should one consider temperature? The case of Epimecis hortaria F. (Lepidoptera: Geometridae) feeding on Lindera benzoin L. (Lauraceae). Environ Entomol 35:600–606 Norghauer JM, Malcolm JR, Zimmerman BL (2008) Canopy cover mediates interactions between a specialist caterpillar and seedlings of a neotropical tree. J Ecol 96:103–113 Ostfeld RS, Manson RH, Canham CD (1997) EVects of rodents of survival of trees and seedlings invading old Welds. Ecology 78:1531– 1542 Pearson AK, Pearson OP, Gómez IA (1994) Biology of the bamboo Chusquea culeou (Poacea: Bambusoideae) in southern Argentina. Vegetatio 111:93–126 Pearson TRH, Burslem DFRP, Goeriz RE, Dalling JW (2003) Interactions of gap size and herbivory on establishment, growth and survival of three species of Neotropical pioneer trees. J Ecol 91:785–796 RaVaele E, Kitzberger T, Veblen T (2007) Interactive eVects of introduced herbivores and post-Xowering die-oV of bamboos in Patagonian Nothofagus forests. J Veg Sci 18:371–378 Reader RJ (1992) Herbivory as a confounding factor in an experiment measuring competition among plants. Ecology 73:373–376 Reid S, Díaz IA, Armesto JJ, Willson MF (2004) Importance of native bamboo for understory birds in Chilean temperate forests. The Auk 121:515–525

123

Oecologia (2009) 161:771–780 Richards LA, Coley PD (2007) Seasonal and habitat diVerences aVect the impact of food and predation on herbivores: a comparison between gaps and understory of a tropical forest. Oikos 116:31–40 Rousset O, Lepart J (2000) Positive and negative interactions at diVerent life stages of a colonizing species (Quercus humilis). J Ecol 88:401–412 Royo AA, Carson WP (2006) On the formation of dense understory layers in forests worldwide: consequences and implications for forest dynamics, biodiversity, and succession. Can J For Res 36:1345–1362 Royo AA, Carson WP (2008) Direct and indirect eVects of a dense understory on tree seedling recruitment in temperate forests: habitat-mediated predation versus competition. Can J For Res 38:1634–1645 SAS Institute (1996) SAS system for Windows, release 6.12. SAS Institute, Cary Schupp EW (1995) Seed-seedling conXicts, habitat choice, and patterns of plant recruitment. Am J Bot 82:399–409 Silva SI (2005) Trophic position of small mammals in Chile: a review. Rev Chil Hist Nat 78:589–599 Silvertown J (2004) Plant coexistence and the niche. Trends Ecol Evol 19:605–611 Takahashi K (1997) Regeneration and coexistence of two subalpine conifer species in relation to dwarf bamboo in the understorey. J Veg Sci 8:529–536 Taylor AH, Jinyan H, ShiQiang Z (2004) Canopy tree development and undergrowth bamboo dynamics in old-growth Abies-Betula forests in southwestern China: a 12-year study. For Ecol Manage 200:347–360 Thompson L, Thomas CD, Radley JMA, Williamson S, Lawton JH (1993) The eVects of earthworms and snails in a simple plant community. Oecologia 95:171–178 Veblen TT (1982) Growth patterns of Chusquea bamboos in the understorey of Chilean Nothofagus forests and their inXuences in forest dynamics. Bull Torrey Bot Club 109:474–487 Veblen TT (1989) Tree regeneration responses to gaps along a transandean gradient. Ecology 70:541–543 Veblen TT, Donoso C, Kitzberger T, Rebertus A (1996) Ecology of southern Chilean and Argentinean Nothofagus forests. In: Veblen TT, Hill R, Read J (eds) The ecology and biogeography of Nothofagus forests. Yale University Press, New Haven, pp 293–353 Wada N (1993) Dwarf bamboos aVect the regeneration of zoochorous trees by providing habitat to acorn-feeding rodents. Oecologia 94:403–407