Post a new comment on this article
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Why should this posting be reviewed?
See also Guidelines for Comments and Corrections.
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.close
A comment on ‘On Population Growth Near Protected Areas’ (Joppa et al., 2009, PLoS ONE 4:1)
Posted by Andy_Nelson on 03 Feb 2009 at 07:36 GMT
Andrew Nelson(1), Graeme M.Buchanan(2), Andrew J. Tatem(3), Glenn Hyman(4), Uwe Deichmann(5) and Andrew Hartley(6)
Contact Author: Andrew Nelson, firstname.lastname@example.org
Note: AN, GH and UD are the developers of the UNEP datasets that are heavily relied upon in both articles.
Wittemyer et al [1,2] reported increased human population growth in the vicinity of Protected Areas compared to background rural growth. Joppa et al  contradict these findings, demonstrating that the results of Wittemyer et al are artefacts of mixing two non comparable datasets. In addition to shared concerns about the Wittemyer et al analysis, those of us who developed the UNEP data [4,5], upon which the findings of both papers are based, welcome the opportunity to clearly reiterate and clarify the nature of the data. We have five main points.
1. The UNEP data are based on total growth rates.
2. Total growth rates are higher than rural growth rates in the studied countries. National-level total growth rates are higher than national-level rural growth rates in 44 out the 45 studied countries (based on UN data). 259 of the 306 studied park buffer zones intersect administrative units which contain towns or cities that influence the growth rate. This influence cannot be removed from the UNEP data to create a rural growth rate.
3. Rural growth rates cannot be extracted from the UNEP data. Rural growth rates were never used in the construction of the data.
4. Growth rates derived from UNEP data cannot be compared to rural growth rates from any other source. They can only be compared to total rates.
5. Deriving any form of accurate growth rate from the published UNEP data is difficult.
*What are the UNEP data?*
The UNEP datasets for Africa and Latin America and the Caribbean contain estimates of population count and population density at 5km resolution for the period 1960-2000 in 10 year intervals. The data and methods are well documented online [4,5] and have also been discussed in the literature . In all cases, the data are clearly labelled as being based on total population growth rates per administrative unit (i.e. state or province).
*What does this mean?*
The administrative units for which growth rates are available have a large extent, it is simply not possible to obtain historical population counts and population growth rates for small sub-national areas in developing countries. By their nature, the majority of these large administrative units will contain at least one “urban” centre or dominant settlement, surrounded by rural areas. In developing countries, there has been a very strong trend of rapid urban growth and slow, sometimes negative, rural growth . This means that the total growth rate – the growth rate of the urban and rural population combined – is almost always greater than the rural growth rate . These total growth rates are applied to the population totals from census’ to generate population estimates per administrative unit at 10 year intervals. The population is then distributed on a 5km resolution raster using a model of accessibility potential which assumes that population densities increase with proximity to urban centres and transport networks. Critically, although the population density within an administrative unit is allowed to vary from pixel to pixel, the growth rate does not.
The population growth rates for all pixels in the UNEP data are total growth rates (regardless of whether the pixels have been identified as urban or rural using other spatial data , developed after the UNEP datasets were published —data that played no role in the generation of the UNEP datasets). There is no delineation of urban areas and there are no urban growth rates in the UNEP data. No rural or urban population, and no urban boundaries were used as input to the UNEP population estimates.
What this means is that – in general - rural growth rates cannot be extracted from the UNEP data by any form of manipulation.
It is possible, where the UNEP data contain growth rates and historical census data at such a high level of detail, that some administrative units will have no dominant settlement, hence the UNEP growth rate is essentially a rural growth rate. How often does this occur? We made an estimate of the number of studied 10km park buffer zones whose boundaries intersect an administrative unit that did not contain a town or city (from . The administrative units were derived from the UNEP datasets by identifying contiguous regions of equal growth rate between 1960 and 2000. Only 47 of the 306 park buffers intersect administrative units that do not contain a town or city. For the remaining 259, the UNEP growth rates will be influenced by the (generally) higher growth rates in these urban centres , no matter how far the urban centre is from the park. This influence cannot be removed from the UNEP data to create a rural growth rate. The coverage of towns and cities  is not necessarily complete and the effect of higher "urban" growth rates may well be more pervasive.
We now address the use of the UNEP data in the two studies.
*Our assessment of the analyses of the UNEP data*
In short, we support the analytical approach of Joppa et al but we stress that the key reason why the validity of the findings of Wittemyer et al are debatable is because they erroneously assumed that the UNEP data contained rural growth rates. As stated above, the UNEP data lack rural growth rate information, and are based purely on total growth rates which are higher than rural rates in 44 of the 45 countries they studied . As also stated above, the growth rates around most of the parks are influenced by higher urban growth rates and that rural growth rates cannot be extracted from the UNEP data. We suggest this invalidates the Wittemyer et al comparison of the UNEP (total) rates against national level UN rural growth rates. Since this comparison forms the key evidence for the conclusions of Wittemyer et al, we agree with Joppa et al that these findings are the result of comparing incompatible datasets, and therefore need to be treated with caution.
Joppa et al also mistakenly refer to the growth rates in the UNEP data as rural growth rates. Although this is erroneous, we assume that they were replicating the original analysis as best they could. However, the Joppa et al study is internally consistent in its use of data; by using the UNEP data to derive differences in annual population growth at increasing distances from 304 parks, they compared like with like (i.e. they do not compare the UNEP population growth rates with growth rates from another source). Thus Joppa et al demonstrate that when appropriate methods are used on the UNEP data, there is no observable difference between total growth rates around protected areas and total growth rates in the wider countryside across the parks in Africa and Latin America.
Spatial databases, such as those produced by UNEP, represent valuable resources for examining large scale trends. However care must be taken in understanding the constraints of the data used to create them and in interpreting findings resulting from their analysis. Complex spatial data sets are becoming increasingly available to researchers, but with the use of these data come responsibilities relating to appropriate use, without which inaccurate conclusions may be drawn accidentally.
*Why do we not publish growth rates for the UNEP data?*
The data are provided as coarse resolution rasters of population counts and population densities since UNEP does not have the rights to distribute the original population data and administrative boundaries, which in many cases were kindly provided by national census and statistical agencies. Furthermore the modelling of the population data makes it very difficult to reverse engineer the original growth rates from UNEP population data. Conversion from vector to raster formats, changes between cartographic projections, mixed pixels at administrative boundaries and the rounding of population estimates to integer values all mean that it is very hard to determine accurate population growth rates over small areas. Any observed differences in the growth rate within an administrative unit are unintentional artefacts due to the issues noted above.
The Protected Area network is the legally recognised backbone of site based conservation. Effective management of these sites, and the biodiversity they support, is likely to make a invaluable contribution to biodiversity conservation. However, conservation resources are limited and consequently spending should to be based on sound evidence. Therefore, perhaps policy makers should use caution when developing strategies until further research has confirmed that population growth is greater surrounding Protected Areas than elsewhere in the country.
We acknowledge that some detailed studies have shown increased population growth in the vicinity of Protected Areas (as documented by Wittmeyer et al). However, we suggest that due to the nature of the UNEP data - the only available continental-scale, spatially-explicit, historical datasets on population - there is no evidence for a detectable difference in human population growth between areas around Protected Areas and the wider countryside across the abovementioned 45 countries and 306 Protected Areas across Latin America and Africa.
Again, we reiterate that the UNEP data contain total growth rates and they cannot and should not be compared against rural growth rates from whatever source. The UNEP data cannot be used to estimate rural (or urban) growth rates. Rural growth rates cannot be reverse engineered out of this data by any means. The documentation – including a discussion on data quality - and data sources for the UNEP data are available online as detailed in the references.
Although the UNEP data are well documented, there appears to be a need for further explanation on the use and limitations of the the data for the conservation community. Those of us who developed the UNEP datasets have tried to make these points heard in earlier communications to Science, but perhaps we have not been clear in our description of the nature of the growth rates used in the UNEP data which are critical to the findings of Wittemyer et al and which also affect the interpretation of the results in Joppa et al . We hope that this extensive comment will form the basis of an informative and useful debate and that it will be welcomed by all concerned.
(1) Independent Researcher/Consultant. Via Borromeo 9, Apt 7. Angera (VA) Italy I-21021
(2) RSPB, 25 Ravelston Terrace, Edinburgh, EH4 3TP, UK.
(3) Dept of Geography and Emerging Pathogens Institute, University of Florida, Gainesville, Florida, USA.
(4) International Center for Tropical Agriculture. AA 6713, Cali, Colombia.
(5) Development Research Group, The World Bank, MSN MC2-205, 1818 H Street, NW, Washington, DC 20433, USA.
(6) Maplecroft. The Towers, St Stephen's Road, Bath, BA1 5JZ , UK
 Wittemyer G, Elsen P, Bean W, Burton ACO, Brashares J (2008) Accelerated human population growth at protected area edges. Science 321: 123–126.
 Supporting online material for Wittemyer G, Elsen P, Bean W, Burton ACO, Brashares J (2008) Accelerated human population growth at protected area edges. Science 321: 123–126. Available at http://www.sciencemag.org...
 Joppa LN, Loarie SR, Pimm SL (2009) On Population Growth Near Protected Areas. PLoS ONE. 6673–6678. 4:1. Available at http://www.plosone.org/ar...
 UNEP & Center for International Earth Science Information Network (CIESIN). (2004) Africa Population Distribution Database (UN Environment Programme GRID Sioux Falls). Available at http://www.na.unep.net/da...
 UNEP International Center for Tropical Agriculture & World Resource Institute. (2000) Latin America Population Distribution Database (UN Environment Programme GRID Sioux Falls). Available at http://www.na.unep.net/da...
 Balk DL, Deichmann U, Yetman G, Pozzi F, Hay SI, Nelson A. (2006) Determining Global Population Distribution: Methods, Applications and Data. Advances in Parasitology. 62: 117-156.
 Montgomery MR (2008) The Urban Transformation of the Developing World. Science 319: 761-764.
 U.N. Population Division (UNPD), World Urbanization Prospects: The 2007 Revision Population Database (UNPD, New York, 2007); available at http://esa.un.org/unup
 Center for International Earth Science Information Network (CIESIN), Columbia University; International Food Policy Research Institute (IFPRI); The World Bank; and Centro Internacional de Agricultura Tropical (CIAT). (2004) Global Rural-Urban Mapping Project (GRUMP), Alpha Version: Urban Extents. Palisades, NY: Socioeconomic Data and Applications Center (SEDAC), Columbia University. Available at http://sedac.ciesin.colum....
 Center for International Earth Science Information Network (CIESIN), Columbia University; International Food Policy Research Institute (IFPRI); The World Bank; and Centro Internacional de Agricultura Tropical (CIAT). (2004). Global Rural-Urban Mapping Project (GRUMP), Alpha Version: Centroids. Palisades, NY: Socioeconomic Data and Applications Center (SEDAC), Columbia University. Available at http://sedac.ciesin.colum....
RE: A comment on ‘On Population Growth Near Protected Areas’ (Joppa et al., 2009, PLoS ONE 4:1)
LucasJoppa replied to Andy_Nelson on 03 Feb 2009 at 13:52 GMT
We welcome Nelson et al.’s detailed comments. Their detailed discussions of the datasets support our conclusions and provide essential insights into those data and their processing. We take this opportunity to apologize to these authors for our not having the good sense to include them as co-authors of our paper in the first place.