E-Letter responses to:

letters: Michael A. Huston;, Andrew Balmford, Joslin Moore, Thomas Brooks, Neil Burgess, Louis A. Hansen, Jon C. Lovett, Si Tokumine, Paul Williams, F. I. Woodward, and Carsten Rahbek People and Biodiversity in Africa Science 2001; 293: 1591-1592 [Full text]

E-Letters: Submit a response to this article

Published E-Letter responses:

Fine-scale Complementarity Analyses Can Reveal the Extent of Conservation Conflict in Africa

Daniel P. Faith, C. R. Margules   (14 January 2002)

Those Complementarity Analyses Do Not Reveal Extent of Conservation Conflict in Africa

Daniel P. Faith   (10 January 2002)

Complementarity Analyses Reveal Extent of Conservation Conflict in Africa

Joslin Moore   (21 November 2001)

Overlap of Species Richness and Development-Opportunity Does not Imply Conflict

Daniel P. Faith   (19 October 2001)

Fine-scale Complementarity Analyses Can Reveal the Extent of Conservation Conflict in Africa 14 January 2002
Daniel P. Faith, Research Scientist Australian Museum, C. R. Margules Respond to this E-Letter: Re: Fine-scale Complementarity Analyses Can Reveal the Extent of Conservation Conflict in Africa	We agree with Moore et al. that the finer scale of land-use planning in Africa will be critical to reconciling biodiversity conservation and human activity. Indeed, we argue that the broad-scale richness-versus- complementarity lesson is just as critical, and even less appreciated, at this finer scale, where planning may include mixed use. It may be a moot point as to whether or not the African species records suggest pressure for conservation and development in the same place, given that conservation and development already occur in the same place in Africa. Studies are underway that are evaluating the biodiversity contributions of such mixed-use areas. Species -richness measures presently dominate such assessments. One recent African study (1), reporting biodiversity conservation in highly utilized areas, found high species-richness of communal grazing lands as compared with reserves. Other African studies (1) also have reported high species- richness in human-use areas. We believe those studies do not provide compelling evidence for reconciliation of regional biodiversity conservation and human activity in Africa. Overlap of species-richness and development opportunities once again is misleading as evidence about conflict - species richness might simply indicate the presence of many widely distributed, abundant species. Just as the "different places" context requires comparison of complementarity value to opportunity cost, this "same place" context should seek a high consequent complementarity value combined with high realized production/development opportunities. A recent review (2) of approaches for "monitoring of biological diversity" focuses on species-richness of areas; complementarity is not mentioned. More encouraging is a recently proposed system for biodiversity monitoring of protected areas having human activities. It rejects species- richness indices in favor of measures indicating "unique" biodiversity components of areas (3). Opportunities to develop a regional perspective on biodiversity monitoring of mixed-use exist where biodiversity gains are predicted but not yet demonstrated (4, 5). In Namibia, it is claimed that "through conservancies, resources are more carefully managed…biodiversity and the environment in general stand to gain" (6). There is now an opportunity in Namibia to estimate the complementarity contributions of conservancies to regional biodiversity conservation goals. A pilot project, extending the PNG complementarity methods, is underway in north-central Namibia (7). The biodiversity contributions of mixed-use areas will be taken into account in identifying low-cost planning solutions for biodiversity protection. We are asking, "what is the complementarity contribution of mixed-use (relative to protected-area conservation or production)?" Crediting these contributions will allow planners to allocate such mixed use (8), along with dedicated conservation or development, as part of a regional strategy that finds a balance among society's needs. References and Notes 1. C. M. Shackleton, Biol. Conserv. 94, 273 (2000). 2. N. G. Yoccuz et al., Trends Ecol. Evol. 16, 446 (2001). 3. F. Danielsen et al., Biod. Conserv. 9, 1671 (2000). 4. J. R. Spence, Tr. Ecol. Evol. 16, 591 (2001); D. Kleijn et al., Nature 413, 723 (2001). 5. Australia's FATE project (Future of Australia's Threatened Ecosystems; see http://www.amonline.net.au/about and M. Archer, Roy. Zool. Soc. N.S.W. (in press)) will attempt such a demonstration; FATE will test through experiments the hypothesis that long-term benefits for biodiversity can be achieved through sustainable use of native resources. Regional biodiversity gains will be assessed by estimating consequent complementarity values. 6. See http://www.dea.met.gov.na/programmes/cbnrm/cons_guide.htm. C. M. Shackleton, Env. Conserv. 28, 270 (2001) argues that conservation of game species has increased within Namibia's conservancies and that such management to conserve key species will benefit many other species. 7. PNG and Namibia conservation planning needs were compared at the recent Zoological Society of Southern Africa Symposium, Port Elizabeth, 9 to 12 July 2001 (P. Barnard, C. R. Margules, D. P. Faith, R. Simmons, "Conservation planning in the real world of land reform, politics, dust and flies"); background on north-central Namibia is available at http://www.dea.met.gov.na/nnep/index.htm 8. For example, as increments in probabilities of persistence of components of biodiversity; D. P. Faith, P. A. Walker, Biod. Conserv. 5, 431 (1996).
Those Complementarity Analyses Do Not Reveal Extent of Conservation Conflict in Africa 10 January 2002
Daniel P. Faith, Research Scientist Australian Museum Respond to this E-Letter: Re: Those Complementarity Analyses Do Not Reveal Extent of Conservation Conflict in Africa	Moore et al.'s defense of their conflict analyses may distract from the general lesson from that dEbate exchange: complementarity, not species richness, best reveals regional biodiversity conservation/human-activity conflict. Moore et al. defend their correlation analyses by arguing that conflict is "likely" because species-rich areas are frequently important in complementarity-based studies. However, most published analyses select "minimum sets" (1). Here, equal "costs" promote selection of species-rich areas. Returning to my example, if equal costs had been assumed instead, species-rich area 4 would form the complementarity-based set. But such findings hardly imply "likely" conflict - whenever species-rich areas are more costly, they may not be selected. Moore et al. say that I characterized their conclusions as not based on least-cost complementarity methods. But I argued that they "estimated the minimum cost…for a set of representative biodiversity areas and cited high total cost as part of their evidence for conflict…they used a simple variant…of the complementarity-cost methods." My key argument remains that "evidence" for conflict based on correlations (fatally) neglects complementarity. Moore et al. say that my concern about inflation of conflict estimates through the use of algorithms without "redundancy" checking is misplaced. This concern was prompted by the absence of this facility in their description of their algorithm. Although redundancy-checking was in fact used, the early variable-cost heuristic algorithms (2) not only depended on redundancy-checking but also variable weights on costs. Their complementarity/cost ratio is akin to using a similar "complementarity minus weighted cost," but with a constant weight of one. Heuristic algorithms limited to using that simple ratio therefore may inflate costs (3). Moore et al. say that "no method…can avoid the fact that in Africa many species simply do not occur in sparsely populated areas." This ignores the important caveat in their original paper - that species sampling records might not reflect true distributions. Balmford et al. noted that future thorough sampling of herptiles having apparently restricted distributions might "resolve some conservation conflicts." Also, their complementarity analysis using only the more evenly sampled taxonomic groups had one-third fewer high-cost cells. That analysis does not evaluate the degree to which apparent restricted-distribution species are sampling artifacts, but does demonstrate high sensitivity of estimated costs to such artefacts. Herptiles and other species may yet prove to be species that "simply do not occur in sparsely populated areas" but, without further assessment of sampling bias, it is unclear whether these species records truly "reveal the extent of conservation conflict" (4). References and Notes 1. R. L. Pressey et al., Tr. Ecol. Evol. 8, 124 (1993); C. R. Margules, R. L. Pressey, Nature 405, 243 (2000). 2. D. P. Faith, P. A. Walker, DIVERSITY: a software package for sampling phylogenetic and environmental diversity 2.1 (1994) (privately distributed). 3. An example illustrates how the complementarity/cost ratio fails to find least-cost sets, whereas variable weighting helps identify least cost sets. Areas I, II, III, and IV are candidate areas for addition to a partial set of areas; species a - g are not yet represented. Legend for rows: 1 = Complementary species; 2 = Cost; 3 = Complementary value; 4 = Ratio; 5 = Ratio after selection of area I; 6, 7, 8 = Difference between complementarity and weighted cost, for weights equal to 0.50, 1.00, 2.00 respectively.      I               II               III             IV 1) a,b,c,d      a - g          e,f             g 2)  1             3.5              2               1 3)  4             7                2                1 4)  4.00        2.00          1.00          1.00 5)  none        0.86          1.00          1.00 6)  3.50        5.25          1.00          0.50 7)  3.00        3.50          0.00          0.00 8)  2.00         0.00          -2.00         -1.00 The ratio method selects areas I, III, and IV at a cost of 4 units; the weighted-cost approach (2) identifies area II as a least-cost set, costing only 3.5 units. 4. The use of 1° grid cells creates another potential sampling bias. An easy way to try to increase apparent "conflict" is to use large grid cells - the larger the cells, the more species there are that are restricted to one cell, and the greater the proportion of total population in any given cell (put all of Africa in one big grid cell and representation of all species will require conservation action in a cell having the total population of Africa). Corroboration of a hypothesis of high conflict requires finding improbably-high cost compared to cost values obtained under null models of species distribution and grid cell size.
Complementarity Analyses Reveal Extent of Conservation Conflict in Africa 21 November 2001
Joslin Moore, Post-doctoral researcher University of Cambridge and University of Copenhagen Respond to this E-Letter: Re: Complementarity Analyses Reveal Extent of Conservation Conflict in Africa	Faith (1) suggests that we have not used complementarity methodology in concluding that considerable conservation conflict exists in Africa. However, he focuses on an article (2) written in response to specific criticisms raised about correlations between functional groups and the relation with NPP reported in a larger earlier analysis (3). In this second paper we reinforced our original message that, in Africa, potential conflict between conservation and development cannot be side-stepped by concentrating conservation effort in sparsely populated areas. However, this conclusion was drawn from the earlier analysis that used the principle of complementarity (3). We agree that a correlation between species richness and human density does not mean that conflicts between conservation and development are obligate. However, we do consider that correlations in species richness and human density suggest that possible conflicts are likely to exist, because areas of high species richness are frequently identified as important areas in conservation planning exercises based on complementarity. Furthermore, that conflict likely is further supported by correlations between endemic species and population density, because it is range-restricted species that most constrain the choice of areas in conservation planning. However, it is only through examining the potential for mitigation by using complementarity-based area selection methods that the magnitude of the issue can be determined. Our conclusions are based on the outcomes of the area selection exercises and not simply on the basis of the correlations. We used area selection methods that use complementarity to choose efficient sets of areas to achieve representation (4). We sought a low-cost solution of the kind suggested by Faith’s example by combining these methods with the overall goal of minimizing the density of people affected. Unfortunately, a low-cost solution was not achievable for the African vertebrate data that we examined. Our inability to identify a low- cost solution is not a failure of the methodology but the result of the particular distribution of species and humans in Africa: endemics and humans congregate in similar areas. In our original analysis (3) we showed that of the 162 cells that must be included to achieve full species representation (called irreplaceable because they contain species that can be represented nowhere else), 79 (49%) of them belong to the 25% most densely populated cells. No method or algorithm however sophisticated can avoid the fact that in Africa many species simply do not occur in sparsely populated areas. Furthermore, Faith suggests that our heuristic algorithm will overestimate the costs involved in achieving representation. The rationale is that our algorithm does not take into account the dynamic nature of complementarity since cells that are most efficient for representing species at the outset may become less efficient as other cells are added to the set. The possibility that areas chosen early in a heuristic selection procedure may become redundant (or less efficient) at a later date was recognized some time ago (5), and redundancy checking to take account of this possibility has been incorporated in WORLDMAP algorithms for some time (4). Interestingly, we have recently used C-plex (6) to calculate truly optimal sets of areas that represent all species but minimize total population density in the selected areas. In this particular instance our heuristic solutions are very close to the optimal solution. The optimal and heuristic solution found a set containing the same number of areas and differed only in identity of six of these areas. In addition, the cost of the heuristically derived set was less than 0.05% more than that of the optimal set. Finally, we do not suggest that the conservation conflict identified in our study is impossible to resolve nor that conservation in these areas is impossible. We simply suggest that successful conservation in Africa must incorporate consideration of people’s needs and activities and requires that conservationists seek creative and equitable solutions to resolve land-use conflicts. Faith makes a number of very welcome suggestions as to how conservation strategies could be refined by incorporating people’s needs specifically into priority-setting exercises. We strongly support and advocate such an approach, particularly for fine- scale planning which, though much needed, our analysis does not seek to provide. Joslin Moore1,2, Andrew Balmford1, Thomas Brooks1,2,3, Neil Burgess4, Michael Folkmann5, Louis A. Hansen2, Jakob Krarup5, Jon C. Lovett6, Si Tokumine6, Paul Williams7, F.I. Woodward,8 and Carsten Rahbek2 1 Conservation Biology Group, Department of Zoology, University of Cambridge, Downing Street, Cambridge, CB2 3EJ, UK 2 Zoological Museum, University of Copenhagen, Universitetsparken 15, DK- 2100, Copenhagen Ø, Denmark 3 Center for Applied Biodiversity Science, Conservation International, 1919 M Street NW, Suite 600, Washington, D.C. 20036, USA 4 Wildlife Conservation Society of Tanzania, Pamba House, P.O. Box 312, Morogoro, Tanzania 5 DIKU, Department of Computer Science, University of Copenhagen, Universitetsparken 1, DK-2100 Copenhagen Ø, Denmark 6 Environment Department, University of York, York Y010 5DD, UK. 7 Biogeography and Conservation Laboratory, The Natural History Museum, Cromwell Road, London, SW7 5BD, UK 8 Department of Animal and Plant Sciences, University of Sheffield, S10 2TN, UK. References and Notes 1. D. Faith, Science Online, available at http://www.sciencemag.org/cgi/eletters/293/5535/1591 (2001). 2. M. A. Huston, et al., Science 293, 1591-1592 (2001). 3. A. Balmford, et al., Science 291, 2616-2619 (2001). 4. P. H. Williams, WORLDMAP 4.1. Priority Areas for Biodiversity. (The Natural History Museum, London, UK., 1996). 5. P. Williams, et al., Conservation Biology 10, 155-174 (1996). 6. CPLEX Linear Optimizer 7.0.0 with Mixed Integer & Barrier Solvers. ILOG 1997-2000
Overlap of Species Richness and Development-Opportunity Does not Imply Conflict 19 October 2001
Daniel P. Faith, Research Scientist The Australian Museum Respond to this E-Letter: Re: Overlap of Species Richness and Development-Opportunity Does not Imply Conflict	Balmford et al. (1) report evidence for conflict between conservation and development in sub-Saharan Africa, extending their earlier findings (2) of a positive correlation over areas between vertebrate species richness and human population density. They now report a positive correlation between species richness and human density for separate vertebrate functional groups, and for plant species richness with human density. For another indicator of development, net primary productivity, high values are found to correspond to high plant species richness. Their conclusion is that all these findings point to a "fundamentally important" problem and that workers "must address this if their efforts to maintain Africa's biodiversity are to succeed" (1). This echoes commentaries on their original paper, warning that conservation and development cannot always be done in different places, and that their work "cuts against much of the ethos of the conservation movement that wants to preserve absolutely pristine environments" (3). Such paradigm-changing announcements, while the popular currency of international biodiversity research, naturally beg closer inspection. I argue here that drawing conclusions about conflict based on species richness/development correlations is unwarranted. Correspondence of species-rich and development-opportunity areas simply does not imply any conflict between biodiversity conservation and development that would require their accommodation in the same place. Acceptance of Balmford et al.'s argument would wrongly discourage the pursuit by conservationists of balanced land-use planning solutions based on priority areas dedicated to biodiversity protection. The weakness of Balmford et al.'s argument results from neglect of a fundamental principle of biodiversity planning, "complementarity." The complementarity value of an area is its biodiversity contribution relative to some baseline (4). For example, it is the additional species relative to those already represented in a current set of protected areas. When opportunity costs of conservation (consequent forgone development) are considered in selecting an additional area for biodiversity protection, complementarity, not richness, is compared to cost (5, 6). Species-poor areas might have high complementarity values relative to their cost, and together make up a low-cost set of protected areas. In the example below, the Spearman rank correlation of species richness with opportunity cost is 1.0, but all species can be represented in a set of areas (the first five) having no opportunity cost (rows are different areas, numbers equal costs, and letters designate species). 0 - a 0 - b 0 - c 0 - d 0 - e 0 - a 4 - a, b, c, d, e 3 - b, c, d, e 2 - a, b, c 1 - d, e Thus, even if all species-rich areas in sub-Saharan Africa had very high opportunity cost, this does not on its own exclude the possibility of finding a low-cost, species-rich set of reserves. A recent whole-country rapid biodiversity assessment (7) of Papua New Guinea (PNG) provides a real example of this disjunction between correlation and conflict and also illustrates how difficult it may be to assess the pressures for mixed-use (development and conservation in the same place) from simple overviews of cost/development distributions. The PNG planning project (7) used the minimum-cost priority-setting methods (5, 6) to identify alternative priority sets of areas for biodiversity protection. The primary index of development-opportunity, interpreted as an opportunity cost of conservation, was forestry production value based on timber volume estimates (7). I have reexamined the PNG data, calculating the product-moment correlation between this opportunity cost and the study's surrogate for species richness. The correlation of 0.63 over 4470 land-units is comparable to the high richness-development correlations reported by Balmford et al. (1, 2). Despite that high correlation, a priority set of areas for PNG achieved high biodiversity representation, with a total opportunity cost equal to only about 7% of the country's total estimated logging opportunity (7). Low total cost was found also for indices of population size, agricultural potential, and land use intensity. Although the PNG study did not prescribe any particular form of management for biodiversity protection in these priority areas, their low overlap with development-opportunity areas suggests that conservation and development largely could be carried out in different places. However, the PNG study highlights some pitfalls in using such analyses to assess pressures for mixed use. One pitfall is simply algorithmic. With 4470 areas and many constraints, a heuristic algorithm for finding minimum-cost sets is practical. The method used in PNG iteratively builds up a set of areas, from random starts, by selecting areas whose complementarity value (relative to the set of other selected areas) exceeds its weighted cost (5, 6). Because the complementarity value of a selected area typically decreases as additional areas are selected, an area must be de-selected if its complementarity value becomes 0 or if the area now can be replaced by a cheaper area(s) providing the same complementary species. Otherwise cost estimates could be inflated. Balmford et al. (2) estimated the minimum cost (overlap with human density) for a set of representative biodiversity areas and cited high total cost as part of their evidence for conflict in sub-Saharan Africa. However, they used a simple variant (2, 8) of the complementarity-cost methods (5, 6) that neglects that dynamic nature of complementarity. Their estimated costs consequently were inflated to some unknown degree, producing a bias favoring the hypothesis that a protected-areas system would involve high conflict. A related problem concerns the assumptions and constraints of such analyses. Constraints such as existing reserve systems and degraded land can make a large difference in cost estimates of achieving a biodiversity target (7). Complementarity hotspots (9) are those areas providing biodiversity marginal gains over a wide range of possible scenarios concerning constraints. If these correspond to development-opportunity areas, there may be a case for mixed use. High endemicity areas naturally will be complementarity hotspots, and some work has examined the degree of overlap of these areas with development-opportunity areas (10). In PNG, overlap of complementarity hotspots and other conservation "must-have" areas with high forestry opportunity areas is a small percentage of total opportunity (7). Another hazard in inferring conflict from overlaps is that the index of development-opportunities might be misleading. In the PNG study, the apparent low overlap of priority sets/complementarity hotspots with development-opportunity does not really capture the issues relating to pressures for mixed use. Population overlap for priority areas appeared low, but the reality is that more than 97% of PNG is under customary land tenure, and it has been argued that "there is no prospect of this land being alienated by the state for purposes of conservation" (11). Thus, political/social realities suggest that there is a large "effective population size," reinforcing other arguments (12) that human population variables may imply greater conflict than appearances suggest. This in turn demands a reinterpretation of timber volume as an index of development-opportunity. The low priority-set overlap with forestry opportunity suggests that government could forgo that relatively small amount of associated royalties to allow dedicated conservation in those areas. However, landowners in PNG have asserted their sovereignty over natural resources (11), and clans in those areas may not be so willing to be the ones to sacrifice forestry income. Further, even when compensation for not logging is initially accepted, a "ransom effect" often emerges such that intensive logging then may occur at a later time (13). These realities suggest that even priority areas with apparent small development opportunity may be "must-haves" for development, analogous to complementarity hotspots as must-haves for conservation. Overlap of conservation and development must-haves implies pressure for mixed use. In PNG, community-forestry is one response to this pressure (11). Logging benefits arguably only stay within the local community through community-based forestry, providing an incentive for community-forestry, with associated biodiversity benefits (11). Indeed, planning may preferentially allocate such mixed-use conservation to areas with high, not low, forestry opportunity (7). These considerations suggest alternative strategies for assessment of "conflict" from cost/development distributions in Africa. First, an estimate of minimum conflict may at the outset take into account the capacity for conflict-avoidance, not just among, but also within areas. Here, the planning challenge is to identify the regional allocations of dedicated conservation, mixed use, and development that together maximise net benefits for society (14). Second, the most useful follow-up assessment may be the evaluation of scenarios that could pose conflicts with this capacity for balanced planning. That perspective on conflict is in accord with important new programs in Southern Africa (15), focussing on scenarios of change and their consequences for maintaining capacity for both biodiversity protection and other ecosystem goods and services. References and Notes 1. A. Balmford et al., Science 293, 1591 (2001). 2. A. Balmford et al., Science 291, 2616 (2001). 3. G. Vogel, Science 291, 2529 (2001). 4. R. I. Vane-Wright, C. J. Humphries, P. H. Williams, Biol. Conserv. 55, 235 (1991). 5. D. P. Faith, P. A. Walker, DIVERSITY: a software package for sampling phylogenetic and environmental diversity 2.1 (1994) (privately distributed). 6. D. P. Faith, P. A. Walker, Biod. Conserv. 5, 417 (1996). 7. See articles in Pac. Conserv. Biol. 6(4) (2001); available at http://www.amonline.net.au/systematics/index.htm. 8. Their algorithm is outlined in more detail in Table 1 in P. H.Williams, M. B. Araújo, Proc. R. Soc. Lond. B 267, 1959 (2000). 9. D. P. Faith, P. A. Walker, Biod. Lett. 3, 18 (1996). 10. A. Balmford, A. Long, Nature 372, 623 (1994). 11. P. Chatterton et al., A future for our forests: strategies for community-based forestry and conservation in Papua New Guinea (World Wide Fund for Nature: South Pacific Program, Suva, 2000). 12. R. P. Cincotta, J. Wisnewski, R. Engelman, Nature 404, 990 (2000). 13. C. Filer, N. Sekhran, Loggers, donors and resource owners. Policy that works for forests and people (National Research Institute, Port Moresby and International Institute for Environment and Development, London, 1998). 14. When mixed use is allocated to an area, it may imply some "partial protection" of biodiversity, expressed as a probability of persistence. These methods are reviewed in (7). 15. B. Scholes et al., Southern African Multi-Scale Millennium Ecosystem Assessment (2001); available at: http://www.ma- secretariat.org/en/assessments/southern.africa.htm. 16. I thank C. Margules for discussion.