SEARCH

SEARCH BY CITATION

Keywords:

  • competition;
  • direct and indirect facilitation;
  • interaction index;
  • light availability;
  • Mediterranean woody species;
  • neighbour effects;
  • seedling growth;
  • soil water content

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  1. Better knowledge of plant interactions is essential to understanding plant community dynamics and has broad applications in restoration and forest operations. Yet studies investigating positive and negative interactions, and indirect interactions, are scarce.
  2. We quantified the nature and intensity of plant interactions between target species, neighbours and ground vegetation in an experimental plantation in southern France to detect competition, direct and indirect facilitation and shifts in interactions with time.
  3. Mediterranean oak seedlings (Quercus pubescens and Quercus ilex) were planted as target species within the following neighbourhoods: pine trees Pinus halepensis at two densities, N-fixing shrubs (Coronilla valentina), mixed pines and shrubs, and controls. The ground vegetation was either weeded or unweeded. We computed the relative interaction index (RII) for survival, growth and elongation of the target species in different treatments over 3 years. Light and soil water content were also monitored.
  4. Results showed that the outcome of plant interactions depended on the variable selected to evaluate plant response. Decrease in Quercus pubescens survival and in diameter growth of both Quercus species indicated competition, whereas enhanced Quercus elongation and higher Quercus pubescens height growth in the first year indicated facilitation.
  5. Indirect facilitative interactions also occurred, resulting in the alleviation of competition by neighbours in the unweeded treatment. This result was explained by the reduction of herb competition caused by shade from neighbours.
  6. With time, competitive interactions became dominant, but were strongly modulated by the type of neighbours. Competition was more severe with shrub neighbours than with pine neighbours due to much greater light interception and also higher water uptake.
  7. Synthesis and applications. We show that the outcome of plant interactions is species specific and varies with time and the indicator selected. Facilitation was detected with tree neighbour treatment, but over time interactions clearly shifted to competition. Hence, forest planting operations can benefit from the interactions induced by neighbouring woody vegetation provided that neighbour species are carefully selected and active post-planting management is performed.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Plant ecologists have channelled considerable effort into understanding and quantifying plant interactions in various environmental conditions and among various species. Emphasis was originally placed on negative interactions, that is, competition as the most obvious key process structuring plant communities competing for space and above- and below-ground resources (Grace & Tilman 1990). However, the role played by positive interactions (facilitation) is now widely recognized, particularly in harsh environments (Callaway 2007; Brooker et al. 2008), and has been documented by numerous studies (e.g. Castro et al. 2004; Gómez-Aparicio et al. 2004; Padilla & Pugnaire 2006; Gómez-Aparicio 2009). Facilitative interactions between two species can be direct, through the improvement of microclimatic conditions or resources, or indirect. For instance, in forest communities, adult trees can limit seedling survival and growth through light interception, but can also prevent the development of competing herbaceous species. Indirect facilitation occurs when the negative effect of shading is less than the positive effect of increasing resources due to herb growth limitation (Levine 1999). By contrast, a net competitive effect is observed if the direct negative effect on seedlings is higher than the indirect positive effect on seedlings induced by a release from herb competition (Pages & Michalet 2003). In plant communities, direct and indirect interactions co-occur, and the outcome of positive and negative interactions is difficult to predict as it depends on numerous factors (Callaway & Walker 1997; Armas & Pugnaire 2005; Maestre et al. 2009). Despite the importance of direct and indirect interactions in structuring plant communities, only a few studies have investigated the role of the interacting species (Callaway 2007), their ontogenic stages (Nuñez et al. 2009; Seifan et al. 2010) and the indicators used to evaluate plant responses (Gómez-Aparicio et al. 2005). Moreover, studies combining all these factors to analyse their role and importance in plant interactions are even scarcer (Seifan et al. 2010).

A better understanding and quantification of plant interactions has practical implications in forestry, in particular for restoration activities in the Mediterranean area, where planting operations usually involve introducing target tree species with removal of pre-existing vegetation (i.e. potential neighbours) to reduce competitive interactions. However, an increasing number of studies emphasize the positive role played by this vegetation (mainly shrubs), especially in arid or semi-arid habitats (Flores & Jurado 2003; Gómez-Aparicio et al. 2004; Padilla & Pugnaire 2006). In more humid Mediterranean conditions, the role of neighbours is more controversial, with competition being thought to play the dominant role. In these areas, introduced seedlings usually undergo stronger competition from herbaceous species, to the detriment of their survival and growth, particularly in oak species (Rey Benayas et al. 2005; Prévosto et al. 2011). To suppress competition by weeds, the most frequently used methods involve different types of treatments, such as herbicide application, mechanical treatments or mulching (e.g. Navarro Cerrillo et al. 2005). However, we hypothesize that the use of neighbours can be advantageous in some situations for the growth and survival of the target species due to indirect facilitation by the limitation of weed development, and so can offer an alternative to less environmentally friendly treatments (e.g. herbicides).

To gain a deeper insight into the effects of neighbouring vegetation on introduced target plants, an experimental plantation was set up with neighbours of different life-forms, that is, trees and/or shrubs, target plants with contrasting life-habits represented by two co-occurring Mediterranean oak tree species (Quercus ilex and Quercus pubescens), and herbaceous species forming the ground vegetation. We explicitly manipulated these three components to disentangle direct and indirect interactions, and we studied the influence of neighbours on target species development and main resource availability over time.

Based on this experiment, we addressed the following more specific issues:

  • What is the relative importance of direct/indirect and of positive/negative interactions between neighbours and target species, and what is the net outcome of these interactions?
  • Do interactions change with time, and do they depend on the identity of neighbours and target species?
  • Do the outcomes of these various interactions depend on the selected plant response variable?

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Site description and species used

The experimental plantation was located in south-eastern France near the town of Avignon (43°54′01″–4°44′55″) in a previously abandoned field. It should be noted that an important rural exodus, in the north part of Mediterranean basin, has led to the reforestation of abandoned agricultural lands. This change in forest land cover is currently progressing on a large scale through secondary succession. Previous agricultural lands usually present rich soils and are of interest to foresters for planting operations. The experimental site used for this study was at an altitude of 80 m a.s.l. Mean annual temperature was 14 °C and mean annual rainfall was 689 mm, but during the experiment (2008–2010), the annual rainfall was largely above this level for the first year (942 mm) and close or higher in the following 2 years (672 and 729 mm). The soil showed a homogeneous composition with a loamy–sandy texture, a low stone load and a high depth (> 1 m) and possessed a high water-holding capacity and fertility. In summer 2007, the pre-existing vegetation was mechanically removed and the soil was scarified to obtain a bare soil. Plantation was carried out in February 2008 using 1-year-old plants grown in a local nursery in 1·2-L containers for the oak species and 0·56-L containers for the other species. We used two late-successional oak species with contrasting leaf habit that co-occur in this region as the target species: the evergreen Quercus ilex L. and the winter deciduous Quercus pubescens Wild. These species are largely promoted for both economic and ecological reasons, and silvicultural treatments leading to pure oak or mixed oak–pine stands were encouraged in our study area. For neighbours, we chose the Aleppo pine tree Pinus halepensis Mill. and the shrub Coronilla valentina subsp. glauca. Aleppo pine is a pioneer light-demanding tree widespread in our area and forming stands that are naturally replaced by Q. ilex or Q. pubescens in the course of succession (Quézel & Barbero 1992). The Coronilla shrub is a common N-fixing species capable of rapid growth in open conditions. Shrubs and in particular legume shrubs have been successfully tested as nurse species in many restoration experiments (Kelty 2006; Gómez-Aparicio et al. 2004).

Treatments

We tested the two target species with five neighbour treatments each: pines at low density (Plo), pines at high density (Phi), Coronilla shrubs (Cor), pines and Coronilla shrub in a mixture (P + C) and a control with no neighbours (Cnt). Target species and neighbours were set up on a 2 × 2·5 m plot: 12 oaks were regularly arrayed in 3 lines of 4 oaks per line. Oak seedlings were spaced at 0·5-m intervals in rows 0·5 m apart (Fig. 1). Neighbours were regularly arrayed using 20 pines for treatment (Plo) so that each oak had 4 neighbours, and using 51 plants, either pines or shrubs or alternating pines and shrubs, for treatments Cor, Phi and P + C, respectively (eight neighbours per oak). We set up a line of neighbours around the plot to limit edge effects, using 18 (Plo) or 36 (Phi, Cor, P + C) regularly spaced plants. In the control plot, we also installed 18 oaks around the perimeter but, as in the other treatments, only the 12 central plants were used for subsequent measurements.

image

Figure 1. Plant distribution in plots according to treatment. Black points indicate target species (Quercus ilex or Quercus pubescens) and white points indicate neighbours (pine or shrub). (a) Pine neighbours (Plo). (b) Pine (Phi) or Coronilla shrub (Cor) or alternating pine and shrub (P+C) neighbours. (c) Control.

Download figure to PowerPoint

Two treatments were applied to ground vegetation: herbs were either manually removed twice a year (spring and autumn) or left to grow. Vegetation that naturally developed was composed of diverse weed species, mostly dicotyledonous (we recorded a total of 49 different species in the first year, data not shown).

Treatments were replicated 4 times, giving a total of 80 plots (2 oak species × 5 neighbour treatments × 2 vegetation treatments × 4 replicates). Plots were distributed in eight blocks (25 × 12 m), with each block containing the two target species and the five neighbour treatments randomly distributed within the block. Plots were randomly assigned to the weeding treatments, with half of the plots being manually weeded (four weed controls during the experiment), while the other half were left unweeded. Plots were separated by a distance of 2 m in each block, and a minimum buffer distance of 4 m was left between each block. Lastly, the whole experiment was fenced to prevent any damages by small herbivores. Some views of the experiment are shown in Supporting information (Figs S1–S4).

Measurements

Just before planting, oak seedlings were cut to a height of 10 cm and shrubs were cut at 15 cm high to limit transplant shock. After plantation, stems of all the oak seedlings were tagged at about 2 cm above-ground and measured (mean and standard deviation: Q. pubescens 6·01 mm ± 1·72, Q. ilex 4·53 mm ± 1·54). Stem diameter was taken as an estimator of plant biomass. We had previously established strong linear relationships (data not published) between basal stem diameter and above-ground biomass on naturally established seedlings in the same area [Q. pubescens biomass (g) = −3·98 + 0·84 × diameter (mm) = 23, R= 0·90, P < 0·001, and Q. ilex biomass (g) = −1·07 + 0·44 × diameter (mm), = 20, R= 0·76, P < 0·001]. We did not detect any differences in diameter between the different treatments (= 1·13 and = 0·34, = 1·36 and = 0·25 for Q. pubescens and Q. ilex, respectively). At the end of each growing season (2008, 2009 and 2010), height and stem diameter of all the oak seedlings was measured together with a subset of four neighbours per plot (only height was measured for shrubs). In addition, in the last year, ground vegetation was collected on four 25 × 25 cm quadrats per plot (unweeded treatment only) and a total of 40 plots equally distributed among the five neighbour treatments were sampled. Vegetation was cut, dried (72 hours at 60 °C) and weighed to obtain the herb biomass per surface unit.

Soil water content (SWC) was measured at different time intervals during the 2009 and 2010 growing seasons using a TDR profile probe (PR2; Delta-T Devices, Cambridge, UK) equipped with five electronic sensors arrayed at fixed intervals. The probe was inserted in an access tube driven into the soil, and we took the average of three readings at each location by rotating the probe through 120°. We installed 1 tube in each plot with Q. ilex seedlings (total of 40 tubes). As topsoil layers were disturbed due to plantation, we investigated SWC at only two soil depths in undisturbed layers: −60 and −100 cm.

Photosynthetically active radiation (PAR, 400–700 nm) was measured at three locations per plot using a linear ceptometer (Decagon Devices, Pullman, WA, USA). Measurements were taken simultaneously above and below the plot canopy using an external sensor during clear days in June 2008–2010 (11:00–14:00). For measurement below plot canopy and in the control plots, the ceptometer was positioned 20 cm below the oak apex. Light transmittance for each plot was then computed as the ratio of the mean PAR values measured above and below the canopy plot.

Data analysis

Target species performance in response to neighbours and vegetation management treatments was analysed using plant growth and morphology descriptors and plant survival. Plant growth was described by total plant height and stem base diameter, as the seedlings had the same initial dimensions. Plant morphology was described by the ratio of height to stem diameter, a ratio widely used in forestry related to plant elongation or slenderness.

Relative neighbour interaction intensity was quantified using the relative interaction index (RII) proposed by Armas, Ordiales & Pugnaire (2004), which is symmetrical around zero, ranges from −1 (only competition) to + 1 (only facilitation), and has powerful statistical properties.

For plants growing in the presence of an herb layer, the index is expressed as:

  • display math(eqn 1)

and without herbs as

  • display math(eqn 2)

where D is the plant descriptor (status: alive or dead, height, diameter, height–diameter ratio) with (n) or without (o) neighbours and in presence (h) or in absence (o) of herbs. Thus, for instance, D(n,h) refers to the target plant growing with neighbours and with herbs, while D(o,o) refers to the target plant without neighbours and herbs. Indices were computed each year in the weeded and unweeded treatments at plot level for each individual using:

  • display math(eqn 3)

where RIIi,j is the index for individual i, in treatment j, j ∈ {Plo, Phi, Cor, P + C} D•,Cnt is the mean of all individuals of the control treatment (in the weeded or unweeded treatment).

To produce an index with mean = 0 in the control treatment, we added a corrected term to Equation 3 leading to the following index, used hereafter:

  • display math(eqn 4)

The index was computed for each living target plant, and the mean was then established at plot level.

Influences of the neighbour and vegetation treatments on seedling diameter, height and height : diameter ratio of the last year for both species were analysed using a mixed model (GLM) to take into account the split-plot design of the experiment. Plots were considered as a random factor and treatments as fixed factors. Comparisons between treatments were analysed using Tukey post hoc tests.

To analyse changes of interaction indices and abiotic variables (light, soil moisture) with time and treatments, we produced linear mixed effects models (procedure lme, package nlme, R software) to deal with repeated measures data, that is, data generated by observing a number of individuals repeatedly under differing experimental conditions where the individuals are assumed to constitute a random sample from a population of interest (Laird & Ware 1982). For each target species, we used neighbour treatment, vegetation treatment and time as fixed factors and plots nested in blocks as random factors.

Prior to the analysis, we checked for anova assumptions (normality, homogeneity of variances) and performed mathematical transformations if necessary to meet these conditions. We used log transformations for dimensions (height and diameter) and arcsine square root transformation for ratios data (light).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Survival

RII index for Q. pubescens survival was significantly influenced by neighbour and vegetation treatments and time (Table 1). In contrast, Q. ilex RII index was only influenced by time but not by the treatments.

Table 1. Results of the linear mixed effects model for RII survival index (values for intercept not shown). Percentage of variance explained by the different models (identified by fixed factors) are shown
 dF Q. pubescens Q. ilex
F-valueP-valueF-valueP-value
Neighbour treatment (NT)339.75<0.0011.150.35
Vegetation treatment (VT)1189.79<0.0012.780.11
Time (T)254.07<0.0014.960.01
Neighbour×Vegetation (NT × VT)30.710.550.870.46
Neighbour×Time (NT × T)619.71<0.0010.930.48
Vegetation×Time (V × T)2239.96<0.0012.400.10
% Variance explained/model
Neighbour treatment (NT) 8.0 2.0 
Vegetation treatment (VT) 16.4 4.8 
Time (T) 7.9 2.5 

Full model:

NT + VT + T + NT × VT + NT × T + V × T

 91.1 10.7 

Changes of the RII index with time according to the different treatments (Fig. 2) showed that: (i) survival of neither oak species was impacted in the first year, irrespective of neighbour treatment. (ii) Mortality occurred in the last 2 years in the neighbour treatments including shrubs, the neighbour treatment of shrubs alone being the most detrimental to survival. (iii) There was a clear difference between the oak species, the deciduous Q. pubescens being subject to much higher mortality than the evergreen Q. ilex. At the end of the experiment, almost all the seedlings survived in the control plots, both weeded and unweeded, whereas 96% and 81% of Q. ilex seedlings survived below the shrub canopy in the weeded and unweeded treatments, respectively (see data in Supporting Information Table S5). These proportions fell to 42% and 21% for Q. pubescens seedlings.

image

Figure 2. Changes in the relative interaction index (RII) of survival with time for each oak species as a function of the neighbour treatment (mean ± SE) in the weeded and unweeded treatments. Treatments: mixed pines and Coronilla shrubs (P + C), Coronilla shrubs (Cor), high-density pines (Phi), low-density pines (Plo).

Download figure to PowerPoint

Growth

Neighbours developed faster than the oak seedlings regardless of treatment (data shown in Supporting Information Table S6). After 3 years, mean pine height was higher in the weeded treatment than in the unweeded treatment (172·6 cm SD = 21·2 cm and 163·1 cm SD = 18·5 cm, respectively, = 50·4, P < 0·001), whereas no difference was recorded between the two pine treatments (F = 0·05, P = 0·82). Coronilla shrubs reached a lower height than pines (137·7 cm SD = 9·9 cm), and we did not detect any differences in height between the two neighbour treatments with shrubs (F = 2·00, P = 0·17) or between the two vegetation treatments (F = 1·55, P = 0·22). However, Coronilla plants developed larger crowns than pines and covered all plots from the second growing season even in the mixed pine and shrub neighbour treatments where pines were clearly out-competed (data not shown). The mean dry herb biomass per surface unit sharply increased from treatments with shrubs (P + C, Cor, mean < 10 ± 0·54 g m−2) to high-density pines (mean = 78·4 ± 16·4 g m−2), low-density pines (mean = 151·3 ± 18·3 g m−2) and control (mean = 400·7 ± 28·7 g m−2).

Results of the anova showed significant effects (< 0·05) of oak species, vegetation and neighbour treatments on target seedlings dimensions for the last year, whereas height:diameter ratio was only significantly influenced by species identity and neighbour treatments (data not shown). Q. pubescens seedlings were usually higher than Q. ilex seedlings in the control and pine neighbour treatments (Table 2), but the two oaks had similar sizes in shrub treatments. Stem diameter for both oak species decreased from the control to pine neighbours (Plo and Phi) and was smallest in shrub neighbourhoods (Cor or P + C). Diameter was greater in the weeded treatment than in the unweeded treatment except in the shrub neighbours (Cor and P + C). Height followed the same pattern as diameter except that height peaked in the high pine density neighbourhood (Phi) in the unweeded treatment. Seedlings were slender (tall and thin) in the presence of a neighbour compared with the control. In the weeded treatment, slenderness was maximal in the (P + C) and (Plo) neighbourhoods for Q. pubescens and Q. ilex, respectively, and maximal in the (Phi) neighbourhood in the unweeded treatment.

Table 2. Mean morphological indicator (SD in brackets) after 3 years for the two oak species according to the neighbour and vegetation treatments
  WeededUnweeded
D (mm)H (cm)H/D (mm/cm)D (mm)H (cm)H/D (mm/cm)
  1. Letters indicate significant differences between neighbour treatments (Tukey test).

  2. Treatments: control with no neighbours (Cnt), low-density pines (Plo), high-density pines (Phi), Coronilla shrubs (Cor), mixed pines and Coronilla shrubs (P + C)

  3. D, stem base diameter; H, total height

Q. pubescens Cnt

19·4 (a)

(0·59)

85·5 (a)

(4·03)

44·6 (a)

(3·09)

12·5 (a)

(0·49)

61·4 (a.b)

(4·14)

51·1 (a)

(3·60)

Plo

13·1 (b)

(0·59)

73·6 (a)

(4·03)

57·1 (b)

(3·09)

10·3 (b)

(0·49)

60·1 (a)

(4·14)

58·9 (a)

(3·60)

Phi

11·2 (b)

(0·59)

70·4 (a)

(4·03)

66·7 (b)

(3·09)

8·8 (b,c)

(0.49)

69·4 (a.b)

(4·14)

79·5 (b)

(3·60)

Cor

7·3 (c)

(0·91)

40·2 (b)

(6·24)

55·6 (a,b)

(4·78)

7·7 (b,c)

(1·07)

35·3 (b)

(9·07)

46·0 (a)

(7·88)

P + C

6·59 (c)

(0·81)

46·4 (b)

(5·58)

70·5 (b)

(4·37)

7·6 (c)

(0·65)

47·6 (b)

(0·65)

62·9 (a,b)

(4·89)

Q. ilex Cnt

18·2 (a)

3·97)

73·1 (a)

(24·42)

40·1 (a)

(9·28)

8·3 (a)

(2·53)

44·3 (a)

(16·88)

56·2 (a)

(23·33)

Plo

10·0 (b)

(2·62)

76·2 (a)

(32·49)

77·3 (b)

(31·67)

7·3 (a,b)

(2·22)

41·3 (a)

(19·07)

58·0 (a)

(21·45)

Phi

7·7 (c)

(2·31)

58·0 (b)

(24·51)

75·8 (b)

(24·64)

6·7 (b)

(1·87)

49·4 (a)

(20·26)

73·1 (b)

(25·19)

Cor

6·4 (c)

(1·45)

8·0 (c)

(18·41)

60·2 (c)

(29·07)

6·3 (b)

(1·58)

40·4 (a)

(19·87)

64·4 (a,b)

(30·15)

P + C

7·2 (c)

(2·25)

46·4 (b,c)

(20·79)

67·6 (b,c)

(27·99)

7·1 (a,b)

(2·23)

42·4 (a)

(17·21)

60·9 (b)

(20·14)

RII index for diameter was significantly influenced by neighbour and vegetation treatments and time for both species (Table 3). RII index for height showed similar results although neighbour treatment was not significant for Q. ilex and the interactions between neighbour and vegetation treatments were never significant for both species. In contrast, the RII index for the height/diameter ratio was only affected by time for both species and the interaction between the neighbour treatment and time.

Table 3. Results of the linear mixed effects models for the stem diameter, height and height/diameter ratio RII indices (values for intercept not shown)
  d.f.RII_diameterRII_heightRII_height/diameter
F-valueP-valueF-valueP-valueF-valueP-value
  1. Percentage of variance explained by the different models (identified by fixed factors) are shown (full model: see Table 1).

  2. NT, neighbour treatment;VT, vegetation treatment; T, time.

Q. pubescens NT311·36<0·0016·380·0031·600·22
VT176·68<0·0014·910·043·710·07
T2119·55<0·0011010·23<0·0015·080·01
NT × VT36·57<0·0011·350·280·240·86
NT × T61·850·107·45<0·0018·22<0·001
VT × T212·22<0·0018·26<0·0010·630·54
Q. ilex NT37·07<0·0011·430·261·250·31
VT1202·85<0·00144·33<0·0014·360·05
T2476·23<0·00111·11<0·00184·90<0·001
NT × VT32·740·072·270·110·610·61
NT × T65·65<0·0017·21<0·0015·50<0·001
VT × T2120·33<0·0010·940·4032·83<0·001
% Variance
Q. pubescens NT 18·2  16·3 3·5 
VT 11·9  3·8 4·4 
T 38·6  30·7 2·5 
Full model 80·1  61·7 26·5 
Q. ilex NT  1·6  1·2 0·4 
VT 44·4  43·5 3·1 
T 30·7  1·5 34·3 
Full model 91·2  86·1 58·1 

Changes in the relative interaction index values for diameter followed similar patterns for both oak species (Fig. 3). In the weeded treatment, clear competitive interactions occurred from the first year, particularly with shrub neighbours (Cor, P + C), whereas no or weak competition was detected for Q. pubescens with pine neighbours. Competition intensity then strongly increased with time, and differences became more pronounced among the neighbour treatments, following the order Cnt < Plo < Phi < P + C, Cor. Trends in the unweeded treatment were similar, but interactions were non-existent or slightly positive in the first year, and competition intensity was largely reduced for the following 2 years, in particular with Q. ilex.

image

Figure 3. Relative interaction index (RII) for stem diameter (mean ± SE). Treatments: mixed pines and Coronilla shrubs (P + C), Coronilla shrubs (Cor), high-density pines (Phi), low-density pines (Plo).

Download figure to PowerPoint

Relative competition index values for height indicated a shift to competition with time (Fig. 4) and like diameter competition was more pronounced with shrubs (Cor, P + C) than with pines (Plo, Phi). Facilitative interactions were, however, detected for Q. pubescens in the first year in both weeded and unweeded treatments. Like diameter, interaction intensity showed lower variations in the unweeded than in the weeded treatments.

image

Figure 4. Relative interaction index (RII) for height (mean ± SE). Treatments: mixed pines and Coronilla shrubs (P + C), Coronilla shrubs (Cor), high-density pines (Phi), low-density pines (Plo).

Download figure to PowerPoint

Neighbour treatments positively influenced oak slenderness in both species in almost all the treatments (Fig. 5). However, in the unweeded treatment, this influence was less marked for Q. ilex and with shrub neighbours for both species. As shown in Table 3, we did not detect a significant effect of the neighbour treatment.

image

Figure 5. Relative interaction index (RII) for height/diameter ratio (mean ± SE). Treatments: mixed pines and Coronilla shrubs (P + C), Coronilla shrubs (Cor), high-density pines (Phi), low-density pines (Plo).

Download figure to PowerPoint

Light and soil water content

Light transmittance was significantly influenced by the neighbour treatment (F = 901, < 0·001) and time (F = 609 < 0·001) but not by the vegetation treatment (F = 6·9, = 0·05). This parameter decreased with time due to plant neighbour development (Fig. 6). However, while with pine, light reduction remained moderate for the two-first years, with shrubs it fell sharply from the second year as shrubs formed a closer cover (light transmittance values < 8%).

image

Figure 6. Light transmittance (mean + SD) ratio as function of the neighbour treatments for years 2008–2010. Data are shown for the weeded and unweeded treatments. Letters indicate significant differences (P < 0·05, Tukey test) between neighbour treatments for a given year. Treatments: control with no neighbours (Cnt), low-density pines (Plo), high-density pines (Phi), Coronilla shrubs (Cor), mixed pines and Coronilla shrubs (P + C).

Download figure to PowerPoint

Soil water content at −60cm was significantly influenced by the neighbour treatment (F = 17·3 < 0·001), the vegetation treatment (F = 12·7, P < 0·001) and time (F = 327, < 0·001), whereas at −100 cm only, the neighbour treatment and time were significant (respectively F = 6·6, < 0·001 and F = 465, < 0·001).

Soil water content decreased following the order Plo > Phi > Cor, P+C in both vegetation treatments (Fig. 7). For the control, soil moisture was higher (at −100 cm depth) or similar (at −60 cm depth) compared with that of the pine neighbourhoods in the weeded treatment, while in the unweeded plots, the presence of a ground vegetation resulted in lower soil water content than in the pine neighbourhoods (−60 cm layer).

image

Figure 7. Mean soil water content for the different neighbour treatments during two consecutive growing seasons for weeded and unweeded treatments. Treatments: control with no neighbours (Cnt), low-density pines (Plo), high-density pines (Phi), Coronilla shrubs (Cor), mixed pines and Coronilla shrubs (P + C).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

A key role of competition and a moderate role of indirect facilitation

Plant interaction analysis shows that competition is the key process driving interactions among species in this Mediterranean tree plantation. Our results clearly indicate that survival and stem diameter (an indicator of plant biomass) were reduced by neighbour presence for both target species at early stages of tree plantation. RII values for diameter clearly shift from neutral or slightly negative values towards high negative values. At first sight, these results seem to conflict with numerous studies that show that competitive interactions are more likely in mild environment, whereas frequency and importance of facilitative interactions dominate in harsh environments, as usually prevail in the Mediterranean area (e.g. Brooker & Callaghan 1998; Choler, Michalet & Callaway 2001; Callaway 2007). In such conditions, conservation or installation of the neighbouring vegetation was shown to benefit the target species (Castro, Zamora & Hódar 2006; Padilla & Pugnaire 2006; Gómez-Aparicio 2009). This process is largely explained by the mitigation of abiotic stresses by the neighbours, in particular, reduction of water stress and direct radiation (Castro et al. 2004; Gómez-Aparicio, Valladares & Zamora 2006; Padilla & Pugnaire 2006). In our study, abiotic conditions were quite favourable to plant growth: the first year was a particularly wet year, largely above the average, and no unusual droughts occurred in the following years. These mild climatic conditions along with the good soil conditions were beneficial to growth and favoured the development of competitive interactions among plants. Mediterranean ecosystems are submitted to great climatic irregularities as drier and milder years can both occur and are unpredictable. Changes in rainfall amounts can make interactions among plants shift from facilitation to competition and vice versa (Padilla & Pugnaire 2006). Former studies have shown that, even on a same site, facilitation can be detected in dry years, whereas competition was dominant during wet years, emphasizing the importance of environmental conditions on plant interactions outcome (Armas & Pugnaire 2005; Maestre et al. 2009; Gómez-Aparicio 2009).

The experimental design allowed us to assess the indirect impact of neighbourhood on target seedlings, as for each neighbour treatment, the herbaceous ground vegetation was either removed or left. When examining growth response (see Figs 3 and 4), we found competitive interactions were considerably alleviated in the unweeded treatment (the control was also subject to herb competition) compared to the weeded treatment (with a control used without herb competition). An indirect facilitation (sensu Levine 1999) was even detected for Q. pubescens height growth with a pine neighbourhood in the weeded treatment, but this effect was slight and occurred only for the first year. Attenuation of competition and facilitation can be explained by the reduction of herb competition due to the shade provided by the neighbours. Herbs mainly act by reducing water availability, as observed in this study and in previous ones (Rey Benayas et al. 2005; Cuesta et al. 2010). Competition for water was shown to be a major interaction between herb species and tree seedlings in water-stressed systems (Ludwig et al. 2004; Cuesta et al. 2010). Hence, indirect facilitation could play a greater role in systems where competition by herbs is a main limiting factor due to a fast development of strongly competing species such as grasses. Pages & Michalet (2003) also observed, in a temperate forest, that direct negative influence of tree neighbours (due to light interception) is offset by indirect positive effects due to reduced herb competition. In Mediterranean conditions, Cuesta et al. (2010) showed that nurse shrubs indirectly facilitated seedlings introduced beneath their canopies by reducing the competitive capacity of herbs although both direct and indirect interactions were influenced by climatic conditions. Similarly, Maestre, Cortina & Bautista (2004) found, in a semi-arid plantation, that seedlings survival and physiological status were improved under pine canopy compared to open conditions due to limitation of the herbaceous understorey by shade.

Influence of species benefactor and beneficiary identity

We found that competitive interactions were greatly modulated by the neighbour treatment. Survival and growth decreased with pine density, and the N-fixing shrub proved to be very competitive towards oak seedlings, particularly Q. pubescens. The faster development of the shrub and its lower crown transparency compared with the pine led to intense competition with the target seedlings, especially for light and to a lesser extent water. The positive influence of the shrub cover, which efficiently limited ground vegetation development as recorded in this study, was likely to induce soil N enrichment (not measured) and was outweighed by its negative influence on light and soil water availability. Previous studies on the effects on biomass production of mixtures combining an N-fixing plant and a non-N-fixing target tree have yielded conflicting results (Parotta 1999; Binkley et al. 2003). The problem of highly competitive N-fixing species in mixed tree plantations has been usually managed through spatial arrangements of the neighbouring species (e.g. grown in blocks or multiple rows) or by deferring their introduction (Binkley et al. 2003; Kelty 2006). The negative impact of shrubs, regardless of association with pine, was, however, largely mitigated for the more shade-tolerant Q. ilex, growth and especially survival being less strongly reduced than for Q. pubescens. Facilitation interactions usually increase in stressful conditions between a benefactor species and a beneficiary species with the stress tolerance of the latter (Maestre et al. 2009) as is the case for Q. ilex compared with Q. pubescens. Combining neighbour and target species that differed widely in several characteristics such as shade tolerance, growth rate, foliar phenology (particularly deciduous vs. evergreen habit) has usually been recommended in mixed tree plantations to limit competitive interactions detrimental to growth and survival (Kelty 2006). Although shrubs have been shown to be better candidates than trees as nurse plants in many situations (Gómez-Aparicio 2009), this result cannot be generalized. Selecting early-successional woody species (e.g. pine), involved in the same succession as the targeted species (e.g. oak), over other species even with promising attributes (e.g. N-fixing shrubs), has also to be considered (Prévosto & Balandier 2007).

Influence of time and the plant response variable selected

In this study, we found that the neighbour treatment effect was clearly dependent on the variable selected to estimate plant response. This finding was also reported by Gómez-Aparicio (2009) in a meta-analysis of studies manipulating interactions among plants for restoration of degraded ecosystems and by Goldberg et al. (1999) in another meta-analysis on competition in terrestrial plants. Growth in diameter was negatively affected by neighbourhood, but the impact on height growth was less marked and facilitative interactions were even recorded for Q. pubescens oak in the first year in pine neighbourhoods. Facilitative interactions became much more obvious for plant elongation measured by the height/diameter ratio. For both species, plant elongation was higher with neighbourhood than without (Fig. 5), although we did not detect a significant influence of the neighbour treatment. Restoration programs have often aimed at selecting neighbourhoods that maximize seedling emergence and survival, even at the expense of growth. The harsh conditions prevailing in degraded ecosystems usually severely limit survival (Gómez-Aparicio 2009). By contrast, in favourable site conditions, forest managers may tend to maximize other plant responses (e.g. tree growth or tree architecture) to enhance production of total biomass or high-quality products. Oak seedlings grown in open conditions tend to develop a large crown, a large number of shoots and to lose apical dominance, resulting in a ‘bushy’ morphology (Mediavilla & Escudero 2010). On the contrary, the presence of neighbours favours plant elongation and limits crown lateral expansion. This well-known morphological response is due to changes in light availability and quality and is traditionally depicted through the ‘shade avoidance syndrome’ (Grime 2001). Neighbour interactions can be thus useful to improve stem quality. Hence, this study illustrates the importance of the plant response variable selected, as opposing results can be obtained according to whether survival, growth or morphological responses are considered.

We found strong evidence for an increase in competition for survival and stem diameter growth over time. This is in line with previous studies on seedling–neighbour interactions which have usually documented a shift in the outcome of interactions from facilitative effects on emergence to competitive effects on survival and biomass (Rousset & Lepart 2000; Miriti 2006; Fayolle, Violle & Navas 2009). Here, we documented shifts from facilitation to competition for Q. pubescens height growth with pine neighbourhood and from competition to facilitation for Q. ilex elongation. Q. pubescens tended to favour a strategy of ‘competitive avoidance’ (Novoplansky 2009) to escape the neighbour shade resulting in enhanced growth height. However, this strategy, recorded in the first year with pine neighbours, became unsustainable when the light resource was too limiting, that is, with shrub neighbours after 1 year and with pine neighbours after 2 years. By contrast, the Q. ilex strategy was based on ‘competitive tolerance’, resulting in small changes in height growth despite the decreased light availability with time.

Conclusions

Contrary to expectations in Mediterranean conditions, our results show that facilitation is not the dominant process in forest planting operation when conditions are not too stressful. Our study also emphasizes the complexity of biotic interactions.

First, positive or negative effects induced by the neighbourhood depend on the plant response variable selected. Thus, we found positive effects of neighbourhood on elongation, a finding that can be important in tree plantations where tree architecture matters.

Second, the nature of the interactions fluctuates widely with time, a result that has been widely demonstrated for annual plants (e.g. Goldberg et al. 2001; Schiffers & Tielbörger 2006) but less frequently for trees (Boyden et al. 2009). In the first year, a pine neighbourhood in unweeded systems did not limit the target seedling biomass and even promoted height growth. With time, the positive or neutral interactions for growth clearly shifted to competition, as also reported in previous studies (Dulohery, Kolka & McKelvin 2000; McNamara et al. 2006). Hence, the use of neighbouring vegetation in plantations is possible, but requires active post-planting management to maintain the benefits of this silvicultural operation. Such management is costly, but has to be considered in the long term for its possible economic advantages (e.g. improvement of wood quality) and ecological benefits linked to the establishment of a mixed stand (Kelty 2006).

Lastly, the neighbour–target plant interactions appear to be species specific. The finding that shade-tolerant species were better adapted to supporting the light-competitive neighbourhood was predictable, but the stronger competition exerted by shrubs than by pine trees was not unexpected: shrubs have usually been shown to be better candidates than trees for nurse use in Mediterranean and arid environments (Flores & Jurado 2003; Gómez-Apparicio 2009). This suggests that life-form alone is not a sufficient criterion for choosing nurse candidates, but that capacity to take up resources is also of importance.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

This study was supported by the French Region Provence-Alpes-Côte d'Azur and French Ministry of Ecology, Sustainable Development and Energy (MEDDE-DEB).

The authors are especially grateful to Roland Estève, Aminata N'Daye, Caroline Piana, Jonathan Baudel and Willy Martin for data collection.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
jpe12000-sup-0001-FigS1.jpgimage/jpg4723KFig. S1. Photograph: General view of the experiment.
jpe12000-sup-0002-FigS2.JPGimage/JPG1738KFig. S2. Photograph: Treatment with Coronilla shrubs.
jpe12000-sup-0003-FigS3.JPGimage/JPG1698KFig. S3. Photograph: Treatment with high pine density.
jpe12000-sup-0004-FigS4.jpgimage/jpg1713KFig. S4. Photograph: Control (with weeding treatment).
jpe12000-sup-0005-TableS5.docWord document78KTable S5. Number and mean dimensions of the oak seedlings for the 3 years according to the neighbour and vegetation treatments.
jpe12000-sup-0006-TableS6.docWord document51KTable S6. Mean dimensions of neighbours according to years and treatments.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.