J. GREENWOOD, 2004 - Birds as biomonitors: principles and practice. In: Anselin, A. (ed.) Bird Numbers 1995, Proceedings of the International Conference and 13th Meeting of the European Bird Census Council, P채rnu, Estonia. Bird Census News 13 (2000):1-10
BIRDS AS BIOMONITORS: PRINCIPLES AND PRACTICE J. J.D. Greenwood ABSTRACT. Long-term surveillance of bird populations has already provided us with valuable scientific insights into avian population dynamics and the negative impact of various man-made factors such as changes in land management. In this paper, I argue that the best bird monitoring programmes have clear objectives, use the data to raise alarms when populations fall below target levels, and collect information on vital rates in order to aid understanding of the causes of changes in population status. Monitoring programmes should be well designed, employing random sampling where possible to minimise bias, and with simple and repeatable methods that encourage high rates of take up by volunteers, ensuring large sample sizes and better precision. Not only essential for bird conservation, such programmes have wider implications in the monitoring of the health of the wider countryside. British Trust for Ornithology, The Nunnery, Thetford, Norfolk IP24 2PU, UK.
INTRODUCTION Much of our knowledge of bird populations comes from long-term surveillance by networks of amateurs operating over large areas, often whole countries. Grey Herons Ardea cinerea in England and Wales provide a striking example (Fig. 1). Here we see the impacts of density-independent factors (hard winters) and density-dependent recovery from such impacts (Mead et al 1979, North 1979). Another volunteer-based survey, the British Common Birds Census, shows that Wren Troglodytes troglodytes populations fluctuate greatly (Marchant et al 1990) and the annual rates of change are clearly density-dependent (Fig. 2). Indeed, half the examples of demographic analyses of Palaearctic-African migrant passerines quoted by Baillie and Peach (1992) come from volunteer-based surveys. Such work has clearly contributed much to our understanding of population dynamics, largely as a by-product of monitoring directed at Figure. 1 Estimated number of Grey heron Ardea cinerea nests in England and Wales, 1928-95. Arrows show unusually hard winters: note their impacts on numbers and the subsequent rapid recoveries (BTO data).
the conservation management of our birds and their habitats. Far from monitoring being a â€˜trivial activityâ€™ (Krebs 1991), it is important for conservation, valuable for science and intellectually demanding. In arguing this thesis, I shall use British examples because I am most familiar with them; similar examples could be drawn from many other European countries; and the principles are generally applicable.
Figure 2. Changes in Wren T. troglodytes populations in British woodlands in relation to population levels (BTO Common Birds Census data; the population levels are index values).
The importance of monitoring stems from manâ€™s impact on the global environment. Not only does Homo sapiens utilise or divert up to 40 % of global potential net primary productivity for himself and his domestic animals (Vitousek et al 1986) (to the detriment of other species), but in doing so he produces pollutants that may be toxic to other organisms, that may profoundly alter habitats by changing nutrient levels, or that may even affect global climates. Many people wish to conserve wildlife in the face of such pressures. For them to succeed they must know what there is to conserve and its importance (e.g. whether a local population is important in a global context); they must know how wildlife populations change, in order to judge if conservation policies are successful; and they must know why populations change, in order to improve conservation management actions. The first two of these needs can be met by simple surveillance (surveys of numbers repeated at intervals) but the third requires true monitoring. This is characterised by: 1. A programme of surveillance that has clear objectives. 2. The assessment of status in relation to desired targets. 3. The raising of alarms when populations depart from the target values to an unacceptable extent. 4. The illumination of the reasons for any failure to achieve the targets. 5. Contributing to the determination of remedial action. 6. Continuation of surveillance, to monitor success of remedial action undertaken. (See also Furness & Greenwood 1993, Saris et al 2003).
Objectives, Targets and Alarms Objectives are what one is trying to do. For example: to provide sufficiently good data on the birds and their habitat to be able to manage waterbird populations on a group of lakes at sustainable levels. Those setting up so-called monitoring schemes often fail to define their objectives or define them too vaguely. (In the example, one would need to define what one meant by ‘sustainable’). Such failure leads to confusion, inefficiency and, often, failure of the scheme to deliver anything useful. Targets (or ‘reference values’, Saris et al 2003) refer to specific requirements within the broad objectives. For example, a target for a particular lake could be mid-winter populations of 2 000 Tufted Duck Aythya fuligula and 5 000 Pochard A. ferina. It is often easy to state targets precisely but less easy to decide what they should be, for this judgement is entirely subjective unless external criteria are imposed. Is the ‘right’ population of Skylarks Alauda arvensis in Britain what it is now, or what we believe it to have been when we were children, 200 years ago, or in 4 000 BC? It is even more difficult to know when it is appropriate to raise alarms. Consider the British Willow Warbler Phylloscopus trochilus population trend, as it appeared in 1989 (Fig. 3). First, there is statistical uncertainty about the extent of the decline during the 1980s, indicated by the confidence limits. Second, even though the particularly steep decline between 1987 and 1988 is statistically significant, is it ecologically important? Third, does the apparently consistent decline during the 1980s represent a real trend or just the chance succession of a number of poor years? Some early ideas about how to address such questions have been presented by Greenwood et al (1995) and we are continuing to develop such work at BTO (Marchant et al 1997). Figure 3. Trends in the British Willow Warbler Phylloscopus trochilus population. Pecked lines indicate confidence limits for the index (continuous line), indexed on a value of 100 for 1989 (BTO Common Birds Census data. Note the log scale).
Whatever is the precise mechanism for raising alarms, it is clear that they depend on the data being analysed and reported. It may seem obvious that data that are gathered but not analysed and reported are useless; nonetheless data-gathering projects have very often been planned without resources being available for analysing the data and reporting the results. In the past, a major bottleneck in the process of data analysis has been computerisation of the data. This can now be circumvented by data being supplied by the observers on optically-readable cards, on computer discs; or on-line. Reporting is important not only for raising alarms with -3-
those responsible for conservation action but also for encouraging the fieldworkers. Their reward is to see what their data reveal about the populations being monitored and to know that their findings are being used.
Understanding and Remedial Action We must not only gather and analyse the data but also be able to interpret the information obtained. Should the populations being monitored attain their target values, then no more is required - except continued monitoring. But if the targets are not attained we need to know why. The information obtained from the monitoring process may enable us to identify the causes of any problems with sufficient certainty that we can recommend remedial action. For example, if populations decline just after a new pesticide has been introduced and only in the places where that pesticide has been used, we would be justified in recommending withdrawal of the pesticide in order to conserve those populations. More often, the causes of the problem will be less obvious and further research will be needed. A well-designed monitoring programme should always, however, provide enough information to point towards some potential causes of the problems being more likely than others, thus enabling the research to be focussed.
Some Problems Concerning Population Surveillance Gannets Morus bassanus are large, conspicuous and breed in a small number of colonies. It is therefore reasonable to try to count their whole population. Wrens are small, relatively inconspicuous and widespread. A complete census of their populations is impossible. Rather, we need to conduct sample surveys, studying a number of representative areas and extrapolating from them to whole populations. What of Mute Swans Cygnus olor? They are large and conspicuous and, although widespread, they are confined to certain well-defined habitats; furthermore, their distribution in particular areas may be fairly well-known to local birdwatchers. It is tempting to believe that a full census is possible. Experience shows, however, that there are inevitably gaps in the survey coverage unless, as with the Gannet, there are very few sites to be covered. It is almost inevitable that such uncontrolled gaps will bias the results, as they will be concentrated in particular regions or particular habitats. Indeed, for Mute Swans in Britain there is good evidence that the gaps tend to be in places where there are no or few birds (unpublished studies by BTO and The Wildfowl & Wetlands Trust). Since the extent of the bias can rarely be assessed, attempted full censuses in which there are gaps are generally less useful than sample surveys in which a relatively small number of sample sites are fully covered. For the results of a sample survey to be unbiased, the sample sites should be randomly distributed. For the results to be precise, there should be many sample sites. Unfortunately, for volunteer-based surveys it is difficult to satisfy both of these requirements simultaneously. The reason is that observers are less willing to visit study sites to which they are directed (especially if they are far from their homes or in uninteresting habitats) than to visit sites that they choose themselves, so a survey in which volunteers are directed to randomly-located study sites may recruit too few participants to provide sufficiently precise results. It may be necessary to compromise, allowing some potential bias in order to achieve sufficient precision. Exactly what to do will depend on particular circumstances. What is important is that, before one settles on the sampling procedure to be adopted, there should be a proper evaluation of the alternatives, in terms of both precision and bias. Ideally, this should involve a pilot study conducted on the study population; if this is impossible, one should at least use evidence based on studies elsewhere or on a similar species. Saris et al (2003) show how experience from a monitoring scheme can be used to redesign and improve it. Fieldwork methods also need proper prior evaluation, to determine which is best, given the objectives of the monitoring programme. It is not necessarily best to choose the method that gives the most precise estimate of numbers at each individual study site. The reason is that the
precision of the national population estimate depends not just on the precision of the estimates for individual sites but also on the number of study sites. Thus, a method that gives a very precise estimate of the population at each site may be very difficult or time-consuming, so that few observers are willing to use it; few sites are therefore studied. It may therefore give less precise national estimates than a method that gives less precise estimates at each site but which, because it is easier, allows more sites to be covered. Gregory and Baillie (1997) and Gregory, Baillie and Bashford (1997) describe how such considerations were addressed in the new British Breeding Birds Survey, while van Strien et al (1994) used the Dutch Broedvogel Monitoring Project to show how one can assess the sensitivity with which a scheme can detect population trends.
Vital Rates Monitoring populations and, in particular, understanding the causes of population change is much enhanced if one measures not just population size but also vital rates (reproduction and survival). For example, the population figures suggest that a long-term decline of British Linnet Carduelis cannabina numbers may recently have been reversed (Fig. 4) but the data from nest records suggest that losses at the incubation stage remain higher than in the early 1970s (Fig. 5), suggesting that the species remains vulnerable (Crick et al 1997). For the Raven Corvus corax, information on changes in the British population size from year to year are poor (see Ratcliffe Figure 4. Trends in Linnet Carduelis cannabina populations in British farmland and woodland (BTO Common Birds Census data).
1997) but nest records alert us to potential problems from the late 1970s onwards (Fig. 6) (Crick et al 1997). The Linnet and Raven nest record data simply supplement information on numbers. In other studies, they have provided deeper insight. Thus the declines in Corn Bunting Miliaria calandra populations in many European countries caused alarms to be raised, triggering further research, which suggested that there were problems of habitat quality and food supply during winter (Donald 1997). It was not clear that problems in the summer were really less significant than those in winter until nest record data were analysed: they showed that mean brood sizes for British Corn Buntings have actually increased during the period of decline (Crick 1997). Results for the Willow Warbler Phylloscopus trochilus provide even more insight (Peach, Crick & Marchant 1995). The species declined by 47 % in southern Britain during 1986-'93 but only by 7 % in northern Britain; data from the Nest Record Scheme show that the differences between regions and temporal variations in breeding output have been small; ringing data from 18 Constant Effort Sites showed a marked decline in annual adult survival rates in the south but no change in the north. This pattern suggests that the problems causing the population decline
Figure 5. Daily nest failure rates at the incubation stage for British Linnets Carduelis cannabina. Points are annual means, with bars indicating standard errors (BTO Nest Record Scheme data).
Figure 6. Brood sizes of British Ravens Corvus corax. Points are annual means, with bars indicating standard errors (BTO Nest Record scheme data).
may be in the non-breeding range (since breeding success is unaffected), specifically in places occupied by the birds that breed in southern Britain. Lapwings Vanellus vanellus have decreased in numbers in many parts of western Europe, including Britain. Analyses of general British ringing recoveries show that, while survival rates vary considerably from year to year (mainly in relation to winter weather), they have not declined during the period of population decline (indeed, they have increased); however, a simple population model shows that Lapwings need to produce 0.83 chicks per pair annually to sustain a stable population; this figure is rarely attained (Peach, Thompson & Coulson 1994). The problem for this species thus appears to be one of declining quality of the breeding habitat. A very similar story could be told for the Grey Partridge P. perdix (Potts & Aebischer 1995).
British Song Thrush Turdus philomelos populations have also decreased since about 1975. Baillie (1990) was able, through a demographic model, to show that the decline could not be explained as a simple effect of winter weather (though this does affect Song Thrush survival). Furthermore, nest record and ringing data showed no change in breeding success or adult survival coincident with the population decline. Baillie suggested therefore that its cause might be a decline in the survival of full-grown birds in their first year of life. Further analysis showed such a decline, from 0.484 during 1962-'75 to 0.405 during 1975â€“'93, a change large enough to explain most of the population decline (Thomson, Baillie & Peach 1997). These later examples show the value of modelling population processes in order to identify the stages of the life cycle at which problems are occurring. Further research can then start to discover the exact nature of those problems. Fig. 7 shows how BTO draws together data from a variety of schemes (based on fieldwork by amateurs) to build population models. These occupy a central position in the programme of Integrated Population Monitoring (Fig. 8). The full range of BTO and other British surveys is described by Greenwood and Carter (2002). Figure 7. How population model can be built using data from various BTO schemes.
Figure 8. The BTOâ€™s Integrated Population Monitoring programme.
Is Monitoring Necessary? It can be argued that, since birds depend on their habitats, conservation requires only that we conserve habitats; so long as we do so, it is argued, the birds will flourish and it is unnecessary to monitor them. The decline of many European raptors during the 1950s and 1960s disproves this. In general, their habitats did not change markedly, the declines being caused by toxic pesticides. Similarly, populations may change in response to changes in food supplies, parasites, predators or competitors, even though the habitat appears stable. Furthermore, they may decline because the habitat changes in their winter quarters or on their migration routes, in places from which little direct information is available. It is essential to monitor the populations themselves. Note that the magnitude of the problems caused by pesticides was not realised in Britain until a survey of Peregrines Falco peregrinus was conducted, in response to complaints from racing-pigeon enthusiasts that Peregrine numbers had unduly increased; the survey showed the converse (Ratcliffe 1993)! If there is no surveillance of populations, problems will probably not be discovered. If surveillance is not continued after actions have been taken to deal with problems, we shall not know whether those actions have been successful. Furthermore, if we restrict ourselves to simple surveillance rather than true monitoring, our appreciation of the problems may be limited because of the absence of clear objectives and targets for the work; and without a system for raising alarms, the surveillance will have only limited practical value. Even if we have objectives, targets and alarm-systems, if our monitoring system is not designed to provide understanding of potential population changes then when such changes are detected we shall not know their likely cause, or how to reverse them. Research that is not begun until we discover that a population is declining will take some time to produce results, by which time the population may have declined to extinction. This is why we need to have monitoring programmes that can deliver some understanding of the causes of population declines at the same time as these are detected, to guide immediate action and focus further research. Longterm, geographically-widespread data can be very illuminating as to the causes of population declines, through the analysis of both temporal and spatial correlations (Green 1995).
The Wider Significance of Monitoring Bird Populations Many species of farmland birds are decreasing in numbers and range in Germany (Flade & Steiof 1990, Bauer & Heine 1992, George 1996), central Europe generally (Bรถhning-Gaese 1992), The Netherlands (Saris et al 1994), and Britain (Fuller et al 1995). Indeed, lowland farmland holds more species with an Unfavourable Conservation Status than any other European habitat and agricultural intensification is the chief threat to European birds (Tucker & Heath 1994), the problems extending over a wide range of farmland types (Pain & Pienkowski 1997). There is very little general evidence on how other taxa are faring on farmland but there can be no doubt, from the bird monitoring, that profound changes have occurred in that habitat and that they are affecting the wildlife in it. Thus, birds can be used to monitor environmental change (Furness & Greenwood 1993, Saris et al 2003). Indeed, in Britain, the monitoring of birds has been important in changing public perceptions and official policies (Greenwood 2003). The monitoring of birds is a particularly cost-effective form of environmental change monitoring because: 1. Birds are easy to observe, identify and count (Sutherland 1996), so more information can be obtained for a given amount of effort than for other taxa. 2. The ecology of birds is better known than that of other taxa, so the results can be more readily interpreted. 3. Birds are high in food-chains, so they may integrate the effects of factors operating at various levels.
4. There are skilled and dedicated work-forces of amateur ornithologists in most countries, so large amounts of data can be collected. It is important that we build on these advantages, to provide a means of monitoring the general health of our European countryside as a habitat not just for birds, but also for wildlife more generally. In doing so, we can build on another advantage of birds over other organisms that some sort of bird monitoring is carried out in a large proportion of European countries (Marchant, Forrest & Greenwood 1998). Conferences such as this one allow the sort of exchange and interaction that has led to constant improvements in survey design, in methods of analysis and interpretation, and in feedback of the results to the conservation authorities. There is the opportunity in every European country, and in Europe as a whole, for us to produce monitoring that makes a major contribution to conservation. It is an opportunity that we should seize.
ACKNOWLEDGEMENTS I thank the tens of thousands of volunteers who make possible the work of BTO and similar organisations throughout Europe. I thank my BTO colleagues for so freely sharing their ideas and for so often, so promptly and so effectively responding to my requests for examples to illustrate particular points. I thank Will Peach and Frank Saris for commenting on an earlier draft. The monitoring work of the BTO is funded by the Trust itself, the Joint Nature Conservation Committee (on behalf of English Nature, the Countryside Council for Wales and Scottish Natural Heritage), and the Royal Society for the Protection of Birds.
REFERENCES Baillie, S.R. 1990. Integrated population monitoring of breeding birds in Britain and Ireland Ibis 132: 151-166. Baillie, S.R., Peach, W.J. 1992. Population limitation in Palaearctic-African migrant passerines. Ibis 134 suppl. 1: 120-132. Bauer, H-G., Heine, G. 1992. Die Entwicklung der Brutvogelbestände am Bodensee: Vergleich halbquantitativer Rasterkartierungen 1980/81 und 1990/91. J. für Orn. 133: 1-22. Böhning-Gaese, K. 1992. Ursachen für Bestandseinbussen europäischer Singvögel: eine Analyse der Fangdaten des Mettnau-Reit-Illmitz Programms. J. für Orn. 133: 413-425. Crick, H.Q.P. 1997. Long-terms trends in corn bunting Miliaria calandra productivity in Britain. In: Donald, P.F. and N.J. Aebischer. (Eds). The Ecology and Conservation of Corn Buntings Miliaria calandra. Joint Nature Conservation Committee. (UK Nature Conservation, No. 13). Peterborough. UK. Crick, H.Q.P., S.R. Baillie, D.E. Balmer, R.I. Bashford, C. Dudley, D.E. Glue, R.D. Gregory, J.H. Marchant, W.J. Peach and A.M. Wilson. 1997. Breeding Birds in the Wider Countryside: their conservation status (1971-1995). BTO Research Report. Donald, P.F. 1997. The corn bunting Miliaria calandra in Britain: a review of current status, patterns of decline and possible causes. In: Donald, P.F. and N.J. Aebischer. (Eds). The Ecology and Conservation of Corn Buntings Miliaria calandra. Joint Nature Conservation Committee. (UK Nature Conservation, No. 13). Peterborough. UK. Flade, M., Steiof, K. 1990. Bestandstrends häufiger norddeutscher Brutvögel 1950-1985: eine Analyse von über 1400 Siedlungsdichte-Untersuchungen. In: van den Elzen, R., Schuchmann, K-L., Schmidt-Koenig, K. (Eds). Current Topics in Avian Biology. J. für Orn. suppl.: 249-260. Fuller, R.J., Gregory, R.D., Gibbons, D.W., Marchant, J.H., Wilson, J.D., Baillie, S.R., Carter, N. 1995. Population declines and range contractions among lowland farmland birds in Britain. Conservation Biology 9: 1425-1441. Furness, R.W., Greenwood, J.J.D. (Eds). 1993. Birds as Monitors of Environmental Change. Chapman & Hall. London. UK. George, K. 1996. Deutsche Landwirtschaft im Spiegel der Vogelwelt. Vogelwelt 117: 187-197. Green, R.E. 1995. Diagnosing causes of bird population declines. Ibis 137 suppl.: S47-S55.
Greenwood, J.J.D., Baillie, S.R., Gregory, R.D., Peach, W.J., Fuller, R.J. 1995. Some new approaches to conservation monitoring of British breeding birds. Ibis 137 suppl.: S16-S28. Greenwood, J.J.D. 2003. The monitoring of British breeding birds: a success story for conservation science? The Science of the Total Environment 310 :221-230. Greenwood, J.J.D & Carter, N. 2003. Organisation eines nationalen Vogelmonitorings durch den British Trust for Ornithology – Erfahrungsbericht aus Groβbritannien. (Organising a national bird monitoring by the BTO-Experiences from Britain.) In: Berichte des Landesamtes für umweltschutz Sachsen-Anhalt. Sonderheft 1/2003: 14-26. Gregory, R.D., Baillie, S.R. 2004. Survey design and sampling strategies for breeding bird monitoring. Bird Census News, vol 13: 19-31 Gregory, R.D., Baillie, S.R., Bashford, R.I. 2003. Monitoring breeding birds in the United Kingdom. Bird Census News, vol 13: 101-112 Hagemeijer, E.J.M., Verstrael, T.J. (Eds). Bird Numbers 1992: Distribution, monitoring and ecological aspects. Proceedings of the 12th International Conference of IBCC and EOAC, Noordwijkerhout, NL. Statistics Netherlands and SOVON. Krebs, C.J. 1991. The experimental paradigm and long-term population studies. Ibis 133 suppl. 1: 3-8. Marchant, J.H., Hudson, R., Carter, S.P., Whittington, P. 1990. Population Trends in British Breeding Birds. British Trust for Ornithology. Tring. UK. Marchant, J.H., Wilson, A.M., Gregory, R.D., Chamberlain, D. and Baillie, S.R. 1997. Enhancements for the monitoring of populations of pest bird species. BTO Research Report. British Trust for Ornithology. Thetford. UK. Marchant, J.H., Forrest, C. Greenwood, J.J.D. 1998. A review of large-scale generic population monitoring schemes in Europe (2nd edn). Bird Census News 10: 42-79. Mead, C.J., North, P.M., Watmough, B.R. 1979. The mortality of British Grey Herons. Bird Study 26: 13-22. North, P.M. 1979. Relating Grey Heron survival rates to winter weather conditions. Bird Study 26: 2328. Pain, D.J., Pienkowski, M.W. 1997. Farming and Birds in Europe. Academic Press. London. UK. Peach, W.J., Crick, H.Q.P., Marchant, J.H. 1995. The demography of the decline in the British willow warbler population. J. of Appl. Stat. 22: 905-922. Peach, W.J., Thompson, P.S., Coulson, J.C. 1994. Annual and long-term variation in the survival rates of British lapwings Vanellus vanellus. J. of Anim. Ecol. 63: 60-70. Potts, G.R., Aebischer, N.J. 1995. Population dynamics of the Grey Partridge Perdix perdix 1793-1993: monitoring, modelling and management. Ibis 137 suppl.: S29-S37. Ratcliffe, D. 1993. The Peregrine Falcon. 2nd edn. T.&A.D. Poyser. London. UK. Ratcliffe, D. 1997. The Raven. T.&A.D. Poyser. London. UK. Saris, F., van Dijk, A.J., Hustings, M.F.H., Lensink, R., van Scharenburg, C.W.M. 1994. Breeding birds in the changing agricultural environment in The Netherlands in the 20th century. In: Hagemeijer, E.J.M. and Verstrael, T.J. (Eds). Bird Numbers 1992: Distribution, monitoring and ecological aspects. Proceedings of the 12th International Conference of EBCC and EOAC, Noordwijkerhout, The Netherlands. Statistics Netherlands (Voorburg/Heerlen) and SOVON (Beek-Ubbergen). pp75-85. Saris, F., F. Hustings, W. Hagemeijer, A. van Dijk, H. Sierdsema and T. Verstrael. 2004. The Dutch breeding bird monitoring scheme: evaluation, new objectives and its merits for conservation. Bird Census News, vol 13: 113-121 Sutherland, W.J. (Ed). 1996. Ecological Census Techniques: a handbook. Cambridge University Press. Cambridge. UK. Thomson, D.L., Baillie, S.R. Peach, W.J. 1997. The demography and age-specific annual survival of song thrushes during periods of population stability and decline. J. Anim. Ecol. 66: 414-424. Tucker, G.M., Heath, M.F. 1994. Birds in Europe: their conservation status. Cambridge: BirdLife International. van Strien, A.J., Hagemeijer, E.J.M., Verstrael, T. 1994. Estimating the probability of detecting trends in breeding birds: often overlooked but necessary. In: Hagemeijer, E.J.M. and T.J. Verstrael. (Eds). Bird Numbers 1992: Distribution, monitoring and ecological aspects. Proceedings of the 12th International Conference of EBCC and EOAC, Noordwijkerhout, The Netherlands. Statistics Netherlands (Voorburg/Heerlen) and SOVON (Beek-Ubbergen). pp525-531. Vitousek, P.M., Ehrlich, P.R., Ehrlich, A.H., Matson, P.A. 1986. Human appropriation of the products of photosynthesis. BioScience 36: 368-373.
- 10 -