Ecological Archives M080-023-A3

Christina M. Kennedy, Peter P. Marra, William F. Fagan, and Maile C. Neel. 2010. Landscape matrix and species traits mediate responses of Neotropical resident birds to forest fragmentation in Jamaica. Ecological Monographs 80:651–669.

Appendix C. Vegetation structure within forest patches by matrix type.

Methods

To capture habitat complexity of forest patches in the four landscape types (forested, agricultural, peri-urban development, and bauxite mining), we measured 12 variables: tree diameter, basal area, canopy height, leaf area index, leaf litter depth, abundance of woody and herbaceous vines, and percent herbaceous cover (0–0.5-m height class), low shrub (0.5–2 m), tall shrub (2–6 m), trees (>6 m), and open canopy. Surveys were conducted at each bird point-count station (with the exception of two bauxite patches due to logistical reasons), totaling 283 vegetation surveys (87 in forest, 77 in agricultural, 49 in peri-urban, and 70 in bauxite landscapes). We established 10-m radius plots centered at each station, divided into four equal quadrats based on 10-m transects in each cardinal direction. We measured tree basal area using a 10-factor prism at each plot center. Within each quadrat, we measured diameter breast height and canopy height of the largest tree, and we estimated the abundance of herbaceous and woody vines based on a categorical scale (0 = absent, 1 = solitary, 2 = few, scattered individuals, 3 = common, 4 = abundant). At 2-m intervals along each transect, we measured leaf litter depth and foliage structure. Foliage structure was scored based on foliage touches along a pole at four height classes (0–0.5 m, 0.5–2 m, 2–6 m, >6 m), with percent cover calculated as the percent of all points at a given height interval with >1 touch (after Schemske and Brokaw 1981).  Percent canopy openness and effective leaf area were estimated based on hemispherical canopy photographs taken at 5-m intervals in two randomly selected transects with a Nikon Coolpix 950 Camera and FC-E8 Nikon Fisheye lens. Percent canopy openness and effective leaf area (Leaf Area Index Ring 4) (Stenberg et al. 1994) were calculated using the Gap Light Analyzer (GLA) program (v 2.0) (Frazer et al. 1999). Sampling was conducted 27 April to 2 July 2005, with all measurements calibrated between two observers. We determined patch-level vegetation by averaging repeat measurements across quadrats, transects, and plots within a patch (N = 97).

Statistical analyses

Due to violations of multivariate normality, differences in overall forest structure by matrix type were determined based on multi-response permutation procedure (MRPP) (with mean standardized variables, Euclidean distance measure, and 1000 permutations) (Mielke and Berry 2001, McCune and Grace 2002) in the R ‘vegan’ package (v 1.13-1) (Oksanen et al. 2008). We conducted MRPP on 10 of 12 vegetation variables, excluding leaf area index and tree diameter due to high inter-correlation (r > 0.6) with percent open canopy and tree height, respectively.

We tested for differences in individual vegetation variables using one-way ANOVAs in the R ‘stats’ package (v 2.8.1) (R Development Core Team 2008). All variables were tested for normality and homogeneity of variances. Familywise error was controlled at α = 0.05 for pairwise comparisons in both MRPP and one-way ANOVAs using the Bonferroni method (Sokal and Rohlf 1995). Tukey’s multiple comparison procedure was used to separate treatment means (Westfall and Young 1993) via the R ‘multcomp’ package (Hothorn et al. 2008). Untransformed means ± 1 standard error are reported.

Results

Forest structure differed by landscape matrix type (MRPP, A = 0.0538, P < 0.0001). Based on pairwise comparisons, fragments embedded in an agricultural matrix differed significantly in forest structure from the other matrix types (forest:  A = 0.0505, P < 0.0001; peri-urban: A = 0.0399, P = 0.0004; bauxite: A = 0.0713, P < 0.0001). Vegetation structure of sites within forested landscapes differed from forest patches in a bauxite matrix (A = 0.0338, P = 0.0003), and marginally from those in a peri-urban matrix (A = 0.0127, P = 0.043). Forest structure of peri-urban and bauxite patches did not differ (A = 0.0072, P = 0.1348).

Five vegetation variables significantly differed among matrix types: tree basal area, tree diameter, canopy height, percent tree layer, and percent herbaceous layer (based on familywise α = 0.05) (Table C1). Tree basal area was significantly higher in patches in an agricultural matrix as compared to patches in forested and bauxite matrices, but did not differ from patches in peri-urban landscapes (ANOVA, F3,93 = 4.84, P = 0.0036). Tree diameter was greater in agricultural patches as compared to all other matrix types (F3,93 = 8.49, P < 0.0001). Canopy height did not differ between patches in agricultural and forested landscapes, but was higher in agricultural patches relative to peri-urban or bauxite patches (F3,93 = 8.13, P < 0.0001). Forest in agricultural landscapes had a greater proportion of trees (>6 m) relative to bauxite landscapes, with no significant difference among patches in forested, agricultural, and peri-urban landscapes (F3,93 = 9.03, P < 0.0031). Percent herbaceous cover was higher in forest patches in bauxite matrices as compared to those in agricultural and forested matrices, but did not differ from patches in peri-urban landscapes (F3,93 = 9.93, P < 0.0001).

Leaf area index, percent open canopy, and percent low shrub and tall shrub layers marginally varied among landscape types at P < 0.05, whereas leaf litter depth and abundance of herbaceous and woody vines did not. Forested landscapes had greater leaf area index (F3,93 = 3.15, P < 0.0286), more tall shrubs (F3,93 = 3.76, P < 0.0144), and less open canopy (F3,93 = 3.8772, P < 0.0286). Forest in bauxite landscapes, and to a lesser extent in peri-urban landscapes, had more low shrubs (F3,93 = 4.03, P < 0.0096).

Overall, forest patches in agricultural landscapes, and to a lesser extent sites within continuous forest, had greater stand basal area, leaf area index, tree diameter, tree canopy height, and tree cover than patches in peri-urban and bauxite landscapes. In contrast, forest fragments in bauxite and peri-urban landscapes had more open canopy and greater proportion of herbaceous cover and low shrubs, indicating these sites may be in earlier successional stages and/or have undergone greater disturbance.

TABLE C1. Means (± 1 SE) of 12 variables measuring vegetation structure of forest patches in landscapes fragmented by agriculture (N = 22), peri-urban development (N = 19), or bauxite mining (N = 25) or of sites within continuous forest (N = 31) in central Jamaica. P values from one-way ANOVAs are provided (with values significant at familywise α = 0.05 in bold). Letters indicate pairwise comparisons based on posthoc Tukey’s HSD tests (P < 0.05).
 
TableC1

 

LITERATURE CITED

Frazer, G. W., C. D. Canham, and K. P. Lertzman. 1999. Gap Light Analyzer (GLA), Version 2.0: Imaging software to extract canopy structure and gap light transmission indices from true-color fisheye photographs, users manual and program documentation, version 2.0. Simon Fraser University and the Institute of Ecosystem Studies, Burnaby, British Columbia, Canada, and Millbrook, New York, USA.

Hothorn, T., F. Bretz, and P. Westfall. 2008. multcomp: Simultaneous Inference in General Parametric Models. R package version 1.0-0.

McCune, B., and J. B. Grace. 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, Oregon, USA.

Mielke, P. W., and K. J. Berry. 2001. Permutation Methods: A Distance Function Approach. Springer-Verlag, New York, New York, USA.

Oksanen, J., R. Kindt, P. Legendre, B. O'Hara, G. L. Simpson, M. Henry, H. H. Stevens, and H. Wagner. 2008. Vegan: Community Ecology Package. R package version 1.13-1.

R Development Core Team. 2008. R: A language and environment for statistical computing, version 2.7.0. R Foundation for Statistical Computing, Vienna, Austria.

Schemske, D. W., and N. Brokaw. 1981. Treefalls and the distribution of understory birds in a Tropical forest. Ecology 62:938–945.

Sokal, R. R., and F. J. Rohlf. 1995. Biometry: The Principles and Practice of Statistics in Biological Research. Third edition. W. H. Freeman and Company, New York, New York, USA.

Stenberg, P., S. Linder, H. Smolander, and J. Flowerellis. 1994. Performance of the Lai-2000 plant canopy analyzer in estimating leaf-area index of some Scots Pine stands. Tree Physiology 14:981–995.

Westfall, P. H., and S. S. Young. 1993. Resampling-Based Multiple Testing: Examples and Methods for P-Value Adjustment. John Wiley and Sons, New York, New York, USA.


[Back to M080-023]