IMIS | Lifewatch regional portal

You are here


[ report an error in this record ]basket (0): add | show Print this page

A sample of bias, the bias in sampling
Hijbeek, R.; Dahdouh-Guebas, F.; Kairo, J.G.; Schoukens, J.; Koedam, N. (2009). A sample of bias, the bias in sampling, in: Mees, J. et al. (Ed.) Book of abstracts: an overview of marine research in Belgium anno 2009. 10th VLIZ Young Scientists' Day. Special edition at the occasion of 10 years VLIZ. VLIZ Special Publication, 43: pp. 71
In: Mees, J.; Seys, J. (Ed.) (2009). Book of abstracts: an overview of marine research in Belgium anno 2009. 10th VLIZ Young Scientists' Day. Special edition at the occasion of 10 years VLIZ. Vlaams Instituut voor de Zee (VLIZ): Oostende. xiii, 221 pp., more

Available in  Authors 
Document type: Summary

    Sampling methods
    Vegetation types > Forests > Mangrove forests
    ISW, Kenya, Gazi Bay [Marine Regions]

Authors  Top 
  • Hijbeek, R., more
  • Dahdouh-Guebas, F., more
  • Kairo, J.G., more

    Forest inventories for either commercial or scientific purposes require accurate data on density and basal area. Mangrove forests grow worldwide on (sub) tropical seashores in saline environments. Within these forests, plotless sampling methods are typically used for density and basal area estimations. A methodological study was conducted to verify the accuracy of several plotless sampling methods, among which the Point Centered Quarter Method (PCQM). Two approaches were used: one fieldwork approach in which the spatial coordinates of all trees in 4 sites in a mangrove forest near Gazi Bay- Kenya were recorded. One modeling approach in which 6 vegetation dispersion patterns were generated based on the literature. Both approaches resulted in datasets on which the plotless sampling methods were applied in silico (MATLAB 7.7.0) and the estimations compared with the true densities.The results show strong bias for each method, depending on the site and pattern. Overestimations and underestimations were found of more than 80%, which gives density estimations with 5 times fewer trees per area than the real value. In general the results correspond with previous findings from for example Engeman et al. (1994), Steinke and Hennenberg (2006) or White et al. (2008). Additionally in this study an emphasis is placed on PCQM, which is the most often used method within mangrove research, and the zonation pattern occurring in some mangrove forests. Overall, PCQM is in the group of worst performing methods. Zonation gives a larger underestimation in every method.One argument given for using PCQM is the assumed comparison benefit, as other researchers also use this method. This study however shows that both underestimations and overestimations do occur when using PCQM, which gives a double bias when comparing two sites.Care should be taken when using plotless sampling methods. Deviant estimations of resources can generate large errors when valuing ecosystem services. Especially with the potential of the Reducing Emissions and from Deforestation and forest Degradation (REDD) situations can come in which communities receive too little or societies pay too much for carbon sequestration.This study argues for a change in the research methodologies. When plotless sampling methods are a necessity because the forest is difficultly accessible or very sparse, it is recommended to measure the distance to the second individual when using PCQM. Preferably however the Variable Area Transect (VAT) method should be used.

All data in the Integrated Marine Information System (IMIS) is subject to the VLIZ privacy policy Top | Authors