It seems to me that the plant distributions in this paper are explained not so much by the geology as by the collection history of New Guinea. The collection bias is not accounted for i.e. one degree grid cells with higher numbers of species may well also have a higher total number of collections. Indeed all of my collections maps for New Guinea indicate that collections are focussed mainly in the north and west of the island around Lae and Bulolo. Is this cause or effect of the findings from this paper?
In order to draw conclusions which are not heavily affected by sampling bias it must be necessary to correct for this e.g. by dividing number of collections by the number of species to get a value indicating the completeness of the flora for that area.
This metric would serve as an indication of perceived or actual species richness in relation to sampling, where a value of 1 indicates that every collection is a unique record of that species in the cell. Values tending towards 0 indicate that the same species have been collected several times within a cell.
Article Regional levels of biodiversity in New Guinea plants
Dear Thomas,
I think you need to use a rarefaction method to estimate of the number of species that would have been recorded if only a set, constant, number of individuals were sampled from each locality. Rarefaction curves can also be used to estimate the number of species that would have been recorded after x samples.
Dear Thomas,
You could solve this uncertainty making accumulation or rarefaction curves to estimate the percentage of species recorded in each grid cell. Different equations based on collector’s curves have been proposed to estimate the theoretical value of species richness. One of the most applied equations is the Clench function. On the other hand, non-parametric estimators have also begun to be used extensively in the recent years.
Accumulation curves are particularly useful and necessary when you compare data obtained by different methodologies or sampling designs (see more information in Moreno and Halffter 2000, Willott 2001).
Here some examples where sampling biases in species richness are assessed (Bruno et al., 2012; Sánchez-Fernández et al.,2008)
Bruno, D., Sánchez-Fernández, D., Millán, A., Ros, R. M., Sánchez-Gómez, P., & Velasco, J. (2012). Assessing the quality and usefulness of different taxonomic groups inventories in a semiarid Mediterranean region. Biodiversity and conservation, 21(6), 1561-1575.
Moreno CE, Halffter G (2000) Assessing the completeness of bat biodiversity inventories using species accumulation curves. J Appl Ecol 37:149–158
Sánchez-Fernández D, Lobo JM, Abellán P, Ribera I, Millán A (2008) Bias in freshwater biodiversity sampling: the case of Iberian water beetles. Divers Distrib 14:754–762.
Willott SJ (2001) Species accumulation curves and the measure of sampling effort. J Appl Ecol 38:484–486
Dear Thomas,
I concur with the previous suggestions, especially those from Daniel Bruno. I'd only add that the equation-based methods appear to be preferred and perhaps more useful, when the data are appropriate for their use. That is often difficult when using found data like historical records due to differences in collection methods, etc. It might prove informative to compare the results of several appropriate methods of estimating species richness or diversity on your question.
Hello Thomas,
Have you looked at Species Distribution Models? I came across an interesting paper, "Species Distribution Models: Ecological Explanation and Prediction Across Space and Time", by Elith and Leathwick, 2009,
http://www.annualreviews.org/eprint/HWR4cusJrXYCSPZ9sUDj/full/10.1146/annurev.ecolsys.110308.120159
when doing just a little web searching. It does seem to cover much of what you are querying, so it may be worth a shot. The full paper and citations are available. Good luck.
http://www.annualreviews.org/eprint/HWR4cusJrXYCSPZ9sUDj/full/10.1146/annurev.ecolsys.110308.120159
Many thanks everyone for your suggestions.
I've since come a long way in this field, yet still have so much to learn (particularly in the rapidly developing realm of SDMs). For my thesis I used a simple rarefaction method (suggested by Peter and Daniel) which proved very insightful. I think this is what I was trying to get at in my original question. I haven't tried creating an accumulation curve yet but of course I have seen examples in text books and look forward to applying this method.
Thank you Kirk for your suggestion of comparing methods of identifying species richness areas and noting the resulting differences; I agree this would be a very useful exercise and one which I will use as I develop my study. In my case, the data are all historical herbarium collections spanning several decades to the present day.
For my thesis I used Species Distribution Models (as suggested by R. T.) in an attempt to; i) account for and overcome sampling bias and; ii) create a regional picture of species richness from relatively few, often clustered, collections. Of course SDMs have their own pitfalls when dealing with spatially biased data and there is no shortage of literature discussing this.
Two other points which I think it is worth considering when estimating species' ranges in general:
i) Mid-domain effect, and the use of equal area bands (as opposed to equal elevation bands) to compare species richness across an elevation gradient.
Bachman, S., Baker, W. J., Brummitt, N., Dransfield, J. and Moat, J. (2004), Elevational gradients, area and tropical island diversity: an example from the palms of New Guinea. Ecography, 27: 299–310
ii) Surface area as opposed to flat or planimetric area when calculating species range - particularly important in areas of undulating or mountainous topography.
Jenness, J. S. 2004. Calculating landscape surface area from digital elevation models. Wildlife Society Bulletin. 32(3):829-839
Thanks again to everybody who replied!