Ecological Archives E096-020-A2

Carol M. Frost, Raphael K. Didham, Tatyana A. Rand, Guadalupe Peralta, and Jason M. Tylianakis. 2015. Community-level net spillover of natural enemies from managed to natural forest. Ecology 96:193–202. http://dx.doi.org/10.1890/14-0696.1

Appendix B. Generalized linear mixed model tables of coefficients.

Table B1. Coefficients from a GLMM with Poisson errors (and log link function) testing whether forest type predicted the abundance of generalist predators in flight traps at control edges between plantation and native forest over all collections. Sampling effort was included as a fixed effect in the full model, and site was included as a random factor. The coefficients presented are from the best-fitting model following model selection. Significant P values (α ≤ 0.05) are indicated in bold.

Fixed effects

Estimate

Std. Error

z value

Pr(>|z|)

Intercept (From Native)

1.7278

0.5762

3.00

0.0027

Forest (From Plantation)

1.7865

0.3096

5.77

< 0.0001

 

Table B2. Coefficients from a GLMM with Poisson errors (and log link function) testing whether forest type predicted the abundance of specialist predators (parasitoids) in flight traps at control edges between plantation and native forest over all collections. Sampling effort was included as a fixed effect, and site was included as a random factor. The coefficients presented are from the best-fitting model following model selection. Significant P values (α ≤ 0.05) are indicated in bold.

Fixed effects

Estimate

Std. Error

z value

Pr(>|z|)

Intercept (From Native)

5.9660

1.3997

4.26

< 0.0001

Forest (From Plantation)

0.6514

0.1549

4.20

< 0.0001

Sampling Effort

-0.3884

0.2207

-1.76

0.0784

 

Table B3. Coefficients from a GLMM with Poisson errors (and log link function) testing whether herbivore reduction treatment, forest type, or their interaction predicted the abundance of generalist predators in flight traps after the herbivore reduction treatment. Sampling effort was included as a fixed effect in the full model, and forest type, within site, within block were included as random factors. The coefficients presented are from the best-fitting model following model selection. Significant P values (α ≤ 0.05) are indicated in bold.

Fixed effects

Estimate

Std. Error

z value

Pr(>|z|)

Intercept (Control, From Native)

3.9287

2.6322

1.49

0.1356

Treatment (Herbivore reduction)

1.1775

0.7663

1.54

0.1244

Forest (From Plantation)

1.6762

0.2325

7.21

< 0.0001

Sampling effort

-2.8406

1.8021

-1.58

0.1150

Treatment*Forest (Herbivore reduction, From Plantation)

-1.7413

0.3181

-5.47

< 0.0001

 

Table B4. Coefficients from a GLMM with Poisson errors (and log link function) testing whether herbivore reduction treatment, collection, forest type, and their interactions predicted the abundance of specialist predators (parasitoids) in flight traps. Sampling effort was included as a fixed predictor in the full model and forest type, within site, within block were included as random factors. The coefficients presented are from the best-fitting model following model selection. Significant P values (α ≤ 0.05) are indicated in bold.

Fixed effects

Estimate

Std. Error

z value

Pr(>|z|)

Intercept (From Native, Before)

4.6244

0.8875

5.21

< 0.0001

Forest (From Plantation)

0.7846

0.1868

4.20

< 0.0001

Collection (After)

-1.5486

0.4485

-3.45

0.0006

Sampling Effort

-0.8033

0.3174

-2.53

0.0114*

*The negative relationship between sampling effort and abundance of parasitoids in flight traps resulted from sampling the most frequently at times of greatest seasonal parasitoid abundance (i.e., the number of days that traps were open happened to be shortest at the time of greatest parasitoid abundance per trap per day).

 

Table B5. Coefficients from a GLMM with Poisson errors (and log link function) testing whether the abundance of specialist predators (parasitoids) in flight traps was predicted by forest type, herbivore reduction treatment, collection, or any interactions between these variables, including only reared species known to parasitize Lepidoptera at these sites. Sampling effort was included as a fixed predictor in the full model, and forest type, nested within site, nested within block were included as random factors. The coefficients presented are from the best-fitting model following model selection. Significant P values (α ≤ 0.05) are indicated in bold.

Fixed effects

Estimate

Std. Error

z value

Pr(>|z|)

Intercept (From Native, Before)

1.0739

0.2503

4.29

<0.0001

Forest (From Plantation)

0.9201

0.2646

3.48

0.0005

Collection (After)

-0.5962

0.1144

-5.21

<0.0001


[Back to E096-020]