Ecological Archives E093-181-A1

Jean H. Burns and Sharon Y. Strauss. 2012. Effects of competition on phylogenetic signal and phenotypic plasticity in plant function traits. Ecology 93:S126–S137. http://dx.doi.org/10.1890/11-0401.1

Appendix A. Tables with additional methods and analysis for "Effects of competition on phylogenetic signal and phenotypic plasticity in plant functional traits." including descriptions of focal species (Table A1), model selection criteria (Tables A2–A5), and a priori model for pot productivity (Table A6).

TABLE A1. Species used in the species interaction experiment to determine the degree of plasticity in species traits, such as SLA, and the potential effect of such plasticity on phylogenetic signal.

Triplet

Focal Species

Confamilial

1

Erigeron foliosus

Solidago californica

1

Erigeron glaucus

Solidago californica

2

Fragaria chiloensis

Potentilla anserina

2

Fragaria vesca

Potentilla anserina

3

Gnaphalium purpureum

Helenium puberulum

3

Gnaphalium ramosissimum

Helenium puberulum

4

Juncus breweri

Luzula comosa

4

Juncus lesueurii

Luzula comosa

5

Rumex occidentalis

Eriogonum latifolium

5

Rumex salicifolius

Eriogonum latifolium

6

Trifolium barbigerum

Lotus heermannii

6

Trifolium macraei

Lotus heermannii

 

TABLE A2. Model selection for traits, SLA and root:shoot using AICc (Burnham and Anderson 2002) with a phylogenetic generalized least squares model with Gaussian distribution (R Statistics, Version 2.11.0). SLA was square-root transformed, root:shoot and total biomass were natural log transformed. Bold values indicate models chosen by AICc and chosen models differed from less preferred models by at least 2 DAICc. Note that the three-way interaction species × soil × interactor could never be included in models as a result of complete mortality by some species in some interactor/soil treatments; the set of models evaluated never included this three-way interaction. The Brownian motion model of evolution was used to model phylogeny for SLA (DAICc > 2, compared with Grafen, not different from OU) and the Grafen model was chosen for root:shoot ratio (DAICc > 2, compared with Brownian and OU). Because models were conducted using PGLS, we used the categorical taxonomic rank "interactor," rather than the continuous variable "phylogenetic distance to interactor," in order to avoid putting branch lengths from the phylogeny in the model twice.

Total biomass

Species

Soil type

Interactor

Species × Soil type

Species × Interactor

Soil × Interactor

 

      AICc

SLA

***

***

***

***

 

 

**

 

3269.915

 

***

***

***

 

 

**

 

3310.952

***

***

***

***

 

 

 

 

3278.838

 

***

***

***

 

 

 

 

3321.413

***

***

***

 

 

 

 

 

3549.441

 

***

***

 

 

 

 

 

3610.033

***

**

 

ns

 

 

 

 

3714.508

 

***

 

ns

 

 

 

 

3801.727

***

 

***

ns

 

 

 

 

   3732.180

 

 

***

ns

 

 

 

 

3838.222

***

**

 

 

 

 

 

 

3705.424

 

***

 

 

 

 

 

 

3793.302

***

 

***

 

 

 

 

 

3722.970

 

 

***

 

 

 

 

 

3831.268

Root:shoot

***

***

*

 

 

ns

 

2351.504

 

.

***

**

 

 

**

 

2396.836

***

***

*

 

 

 

 

2319.578

 

***

***

 

 

 

 

2366.440

***

***

 

 

 

 

 

2311.416

 

***

***

 

 

 

 

 

2356.137

***

 

ns

 

 

 

 

2500.033

 

***

 

ns

 

 

 

 

2494.181

 

***

*

 

 

 

 

2357.223

 

 

***

 

 

 

 

2404.496

***

 

 

 

 

 

 

2488.399

 

***

 

 

 

 

 

 

2482.175

 

***

 

 

 

 

 

2348.348

 

 

***

 

 

 

 

 

2393.711

ns

 

 

ns

 

 

 

 

2533.336

 

 

 

ns

 

 

 

 

2528.103

*** P < 0.001, ** P < 0.01, * P < 0.05, † 0.10 > P > 0.05, ns P > 0.10

 

TABLE A3. Model selection for trait divergence, with phylogeny modeled using the Grafen model for divergence in SLA and the OU model for divergence in root:shoot ratio, as chosen by AICc. Models with phylogeny were preferred to models without phylogeny (DAICc > 2). Divergence in SLA and root:shoot ratio were natural log transformed. Species × soil type, species × interactor, and soil × interactor could not be tested in the PGLS. Models with higher order interactions that could not be tested (due to mortality or small plant size that precluded SLA measurements) are not included.

Total biomass

Species

Soil type

Interactor

 

AICc

Divergence in SLA

*

***

ns

**

 

727.8457

 

***

ns

**

 

730.9047

*

***

ns

 

 

735.1393

 

***

ns

 

 

737.0484

*

***

 

***

 

719.5184

 

***

 

**

 

723.1253

**

 

ns

***

 

725.3860

 

 

ns

***

 

729.5407

*

***

 

 

 

727.7911

 

***

 

 

 

729.8763

**

 

ns

 

 

737.4662

 

 

ns

 

 

741.2024

**

 

 

***

 

718.8169

 

 

 

***

 

722.7510

Divergence in root:shoot ratio

ns

***

***

***

 

612.4880

 

***

***

***

 

606.5201

ns

***

***

 

 

650.3693

ns

ns

 

ns

 

653.8381

 

ns

 

ns

 

649.2391

ns

 

ns

 

666.3739

 

 

*

ns

 

661.2890

ns

ns

 

 

 

650.1598

 

ns

 

 

 

645.4724

ns

 

 

 

662.7309

 

 

*

 

 

657.6534

ns

 

 

ns

 

713.8320

 

 

 

ns

 

709.0255

*** P < 0.001, ** P < 0.01, * P < 0.05, † 0.10 > P > 0.05, ns P > 0.10

 

TABLE A4. Model selection for trait divergence and plasticity to test the hypothesis that trait divergence as a result of phenotypic plasticity is potentially adaptive (Table 4). In these tests, congener, and conspecific competition treatments were treated separately in analyses to remove trait divergence owing to species' differences. The phylogeny was modeled with the Brownian, Grafen, or OU model, as chosen by AICc. The model with phylogeny was preferred to the model without phylogeny in most cases (DAICc > 2), with exceptions noted below. Models that could not be tested (e.g., due to mortality) are not included.

Species

Divergence in trait

Soil type

Species × divergence

Species × soil

Divergence × soil

Species × divergence × soil

AICc

Trait = SLA, competition = conspecific (Grafen)

***

ns

***

 

 

ns

 

354.3137

 

ns

***

 

 

ns

 

367.0289

***

ns

***

 

 

 

 

342.5930

 

ns

***

 

 

 

 

356.4058

***

ns

 

 

 

 

 

353.4284

 

ns

 

 

 

 

 

363.4151

***

 

***

 

 

 

 

338.1516

 

 

***

 

 

 

 

   352.6533

Trait = SLA, competition = congener (OU)

ns

ns

**

 

 

 

138.1199

 

ns

**

 

 

 

137.2301

ns

ns

**

 

 

 

 

130.5612

 

ns

**

 

 

 

 

134.1935

ns

ns

 

 

 

 

 

130.4537

 

ns

 

 

 

 

 

137.1074

ns

 

**

 

 

 

 

127.2657

 

 

**

 

 

 

 

131.6789

Trait = SLA, competition = confamilial (model without phylogeny preferred)

ns

ns

ns

ns

ns

ns

ns

118.2484

ns

ns

ns

ns

ns

ns

 

210.7398

ns

ns

ns

ns

ns

 

 

107.5414

ns

ns

ns

ns

 

ns

 

107.7843

ns

ns

ns

 

ns

ns

 

106.5978

ns

ns

ns

ns

 

 

 

89.3565

ns

ns

ns

 

ns

 

 

89.3523

ns

ns

 

 

ns

 

89.5861

ns

ns

 

 

 

 

83.4379

 

ns

ns

 

 

 

 

83.1105

ns

 

 

 

 

 

80.2613

 

ns

 

 

 

 

 

81.4828

 

ns

 

 

 

 

77.7841

Trait = root:shoot, competition = conspecific (OU model)

ns

ns

***

 

 

ns

 

268.5244

 

ns

***

 

 

ns

 

280.2765

ns

ns

***

 

 

 

 

256.4279

 

ns

***

 

 

 

 

269.9271

ns

ns

 

 

 

 

 

270.0899

 

ns

 

 

 

 

 

283.8748

ns

 

***

 

 

 

 

249.4001

 

 

***

 

 

 

 

263.4522

Trait = root:shoot, competition = congener (OU)

ns

ns

**

 

 

ns

 

143.9204

 

ns

**

 

 

ns

 

142.7785

ns

ns

**

 

 

 

 

125.8187

 

ns

**

 

 

 

 

129.2172

ns

ns

 

 

 

 

 

126.9778

 

ns

 

 

 

 

 

133.5080

ns

 

***

 

 

 

 

119.3582

 

 

***

 

 

 

 

123.5406

Trait = root:shoot, competition = confamilial (model without phylogeny is preferred)

**

*

ns

ns

ns

ns

ns

-63.8925

**

*

ns

ns

ns

 

 

86.5864

**

*

ns

ns

 

ns

 

85.7054

**

*

ns

 

ns

ns

 

84.9496

**

*

ns

ns

 

 

 

49.7918

**

*

ns

 

ns

 

 

49.2313

**

*

ns

 

 

ns

 

48.8251

 

ns

 

 

ns

 

44.3411

***

ns

ns

 

 

 

 

36.3463

 

ns

 

 

 

 

38.4459

***

ns

 

 

 

 

 

28.6722

 

*

 

 

 

 

 

33.2899

***

 

ns

 

 

 

 

28.0294

 

 

ns

 

 

 

 

36.0336

*** P < 0.001, ** P < 0.01, * P < 0.05, † P < 0.10, ns P > 0.10.

 

TABLE A5. The 10 most probable models ranked by AICc, relating standardized total biomass per pot (natural-log transformed), to fixed treatment effects, including divergence in root:shoot ratio and SLA, which were natural-log transformed for analysis.  Soil × phylogenetic distance was included in model selection, but was not included amongst the top 10 models. An 'X' indicates that the effect was included in the model.

Species

Soil

PD

SLA

RS

S × soil

S × PD

AICc

 Rank

X

X

X

X

X

X

X

428.7

1

X

X

X

X

X

X

 

432.2

2

X

 

X

X

X

 

 

434.1

3

X

 

X

X

X

 

X

434.3

4

X

 

 

 

X

 

 

436.2

5

X

X

X

X

X

 

X

439.4

6

X

X

X

X

X

 

 

439.7

7

X

X

 

 

X

 

 

440.8

8

X

X

 

X

X

 

 

442.1

9

X

X

X

 

X

 

 

444.2

10

 

TABLE A6. A priori model for standardized total biomass (natural-log transformed) as a function of random effects, species, species × soil, species × phylogenetic distance, and the following fixed effects. The a priori model had an AICc of 447.9 > 2 DAICc greater than that of the best model (AICc = 428.7), presented in Table 5 (Table A5).

Source

df

df denom.

F ratio

P value

Phylogenetic distance

1

5

13.56

0.01

Divergence in root:shoot ratio

1

255

10.94

< 0.01

Divergence SLA

1

247

6.08

0.01

Soil type

3

11

0.23

0.88

Phylogenetic distance × soil type

3

90

2.96

0.04

Model adjusted R² = 0.44.

 

LITERATURE CITED

Blomberg, S. P., T. Garland, and A. R. Ives. 2003. Testing for phylogenetic signal in comparative data: Behavioral traits are more labile. Evolution 57:717–745.

Burnham, K. P., and D. Anderson. 2002. Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach. Second edition edition. Springer, New York, New York, USA.


[Back to E093-181]