.tbstract

1"o assess conservation priorities, a means o f measuring the distribution of a much larger part o f overall bio- diversity is needed that will at the same time reduce the colossal sampling problems of exhaustive surveys. One possibility is a 'top-down' taxonomic approach, in which the biodiversity of different areas may be compared using measures based on the number of higher taxa present in each. The advantage o f this approach is that survey costs should be greatly reduced because identification to species level, particularly within the few hyper-rich higher taxa, would be unnecessary. We report that family richness is a good predictor of species richness for a variety of groups and regions, including both British ferns and British butterflies among 100 km x 100 km (10,000 km 2) grid squares, Australian passerine birds among 5° x 5° grid squares (c. 220,000-310,000 km 2) and 10° x 10° grid squares (c. 970,000-1,190,000 km2), and North and Central American bats among grid squares of c. 611,000 km 2. With careful choice of higher-taxon rank, it may be possible to re-deploy effort from taxonomically intensive to taxonomically extensive surveys, in order to estimate the global distribution of a much larger proportion o f overall biodiversity at the same cost.

Keywords: biodiversity, higher taxa, species richness, indicator.

INTRODUCTION

One of the problems conservationists face is how can the most important areas for biodiversity (biological diversity) be identified quickly and cheaply. Biodiversity is not only difficult to define, but basic fine-scale data on the distribution of organisms are sadly lacking and potentially very expensive to acquire in sufficient quantity.

Maintenance of biodiversity has become one of the principal goals of conservation. Myers (1979) extended the existing concern for the more conspicuous and vulnerable species by drawing attention to the growing threat to the great variety of all living organisms.

Inevitably, it will not be possible to save everything

Biological Conservation 0006-3207/94/$07.00 © 1994 Elsevier Science Limited, England. Printed in Great Britain

everywhere, so the conservation of biodiversity will have to be based on priority areas (McNeely et al., 1990; Groombridge, 1992; Reid et al., 1992). Many conflicting values need to be taken into account when assessing area priority (Usher, 1986; Spellerberg, 1992) and all should be made explicit and rational (Morowitz, 1991). Consequently we need to be able to measure the contribution an area makes to the overall pattern of biodiversity (May, 1990), preferably sum- marised in a simple univariate measure.

In practice, biodiversity is commonly measured by counting the number of species (species richness) in an area (Groombridge, 1992) and the turnover of species among areas (Vane-Wright et al., 1991). Even advo- cates of a functional approach to biodiversity at the ecosystem level recommend that measures based on species richness are often appropriate at larger scales (Walker, 1992). Yet, for measuring the overall bio- diversity within an area, it would be exceedingly difficult to enumerate all the organisms at even a single small, temperate locality, let alone for tropical areas encom- passing tens or even thousands of square kilometres.

What is needed is a means of measuring the distribu- tion among areas of a much larger part of overall bio- diversity (Solbrig, 1991; Ehrlich, 1992), by which the sampling problem is reduced to manageable proportions (this larger part of overall biodiversity within an area is referred to here using the term wholesale biodiversity). Biodiversity surveys already take a large proportion of conservation budgets and the demand for them is growing: cost-effectiveness is therefore becoming increasingly important (Burbidge, 1991; Pressey & Bedward, 1991). In the past, it has been suggested that this might be achieved more readily if the greatest overall biodiversity was associated with areas that have high values for certain surrogate attributes that are more easily measured.

Here we consider two familiar surrogates for esti- mating the distribution of overall biodiversity: environ- mental variables and indicator groups. A third approach, the use of higher-taxon richness as a surro- gate for species richness, has received less attention and is explored in a preliminary analysis of data for four well-known groups.

25

20

o

o

0-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1.0 Coefficient

Fig. 1. Correlation coefficients calculated by Pianka and Schall (1981: Table 4) between species richness of groups of birds, reptiles, amphibians and mammals in Australia (161 of the 240 × 240 km grid squares), plotted as a frequency histogram. Comparisons are not independent, as some of the groups overlap in species composition: pairwise comparisons of marsupials, all birds, passerines, non-passerines, insectivo- rous birds, omnivorous birds, and seed-eating birds, versus marsupials, frogs, all reptiles, turtles, snakes, all lizards, non- Ctenotus skinks, dry-adapted lizards, agamids, geckos, and

varanids.

Biodiversity and environmental variables

Ecologists have searched for environmental factors that may limit overall biodiversity (e.g. Rabinovich & Rapoport, 1975; Schall & Pianka, 1978; Pianka & Schall, 1981; Wright, 1983; Currie & Paquin, 1987; Woodward, 1987; Owen, 1988; Turner et al., 1988; Adams, 1989; Adams & Woodward, 1989; Willig & Selcer, 1989; Currie, 1991). The use of characteristics of the physical environment as predictors of overall biodi- versity appears attractive because data for physical variables may already be available or may be relatively inexpensive to acquire. Some expensive verification, or 'ground-truthing', is especially important with this approach in order to ensure that the organisms do actually live in any chosen areas in the numbers predicted.

Unfortunately, many of the relationships that have been detected are severely non-linear, with some levels of biodiversity corresponding to both high and low levels of environmental variables (Currie, 1991). The use of these variables may also be restricted to partic- ular kinds of habitat. For example, among the better candidates, evapo-transpiration (Currie & Paquin, 1987; Currie, 1991) clearly cannot be applied to marine or aquatic habitats.

Biodiversity and indicator groups

Conservationists have sought to identify areas of high overall biodiversity by association with areas of rich- ness for relatively small, but particularly well-sampled, groups of organisms. For the terrestrial environment, these groups have included mammals (Mittermeier, 1988), birds (Bibby et al., 1992), butterflies (Brown, 1991; Kremen, 1992), plants (Cronk, 1988) and tiger beetles (Pearson & Cassola, 1992). The use of selected species groups as indicators of overall biodiversity appears attractive, because if suitable indicator

relationships were shown to exist, sampling for just the selected species might greatly red..uce survey costs.

Birds, butterflies and most of'the other popular indi- cator groups have been surveyed particularly inten- sively over long periods because they have large body sizes and include representatives that are colourful and attractive (yet in all cases some taxa remain poorly known). But although they may be among the most obvious or conspicuous components of biodiversity, they are nonetheless only a very small proportion of overall biodiversity (Southwood, 1978; May, 1988; Barnes, 1989; Gaston, 1991a; Holloway & Stork, 1991; Hammond, 1992). Thus, use of these groups requires bold extrapolations, which need to be supported with good evidence for their reliability.

Unfortunately, areas of particularly high species rich- ness cannot be assumed to coincide among different groups of organisms. Even if some species' assemblages do show patterns of 'nested' composition at a local scale of distribution among habitat islands (Patterson

&Atmar, 1986; Cutler, 1991), this does not hold at the global scale. Lemurs are not indigenous to the New World and marmosets are not indigenous to the Old World, no matter how species-rich the local assemblages. Indeed, different higher taxa often reach maxima of species richness in different areas even within continents. Examples of this are known from North America (Schall & Pianka, 1978; Currie, 1991), South America (Gentry, 1992), Africa (MacKinnon & MacKinnon, 1986) and Australia (Schall & Pianka, 1978; Pianka & Schall, 1981). There may even be nega- tive correlations between the regional species richness of different higher taxa (Fig. 1; Schall & Pianka, 1978). Thus, an indicator relationship should always be demonstrated (Landres et al., 1988; Noss, 1990), and not just assumed. Too often, 'indicator groups' seem to be simply the groups which their advocates work with, no evidence being provided that they indicate anything. At the very least, the results from these and from the other techniques for estimating overall biodiversity ought to be compared.

Biodiversity and higher taxa

Here we consider another possibility, a 'top-down' taxonomic approach, in which biodiversity may be compared among areas using the number of higher taxa from across a much broader range of the organisms recorded in each area. The use of higher- taxon richness as an estimate of wholesale biodiversity appears attractive, because it should be substantially cheaper to identify specimens from survey samples to the level of higher taxon than to the level of species.

Intuitively, a relationship between diversity at differ- ent taxonomic levels appears likely. It is regarded as a good approximation in palaeontology (Sepkoski, 1992) and has been assumed for extant taxa (Salm, 1984). If a relationship can be demonstrated, then it might be possible to use the spatial distribution of higher-taxon richness to estimate the distribution of species richness more widely.

As an analogy, consider different numbers of objects (species) in a series of boxes (higher taxa), which are themselves present in different numbers among a series of rooms (localities). All that is being sought is whether or not there is a relationship between the distribution of boxes (higher taxa) among rooms (localities), and the overall distribution of objects (species) among those rooms (localities). Clearly it is essential to using the relationship that the level of effort to count boxes is not allowed to vary substantially among rooms, or else both the counts of boxes and of objects could simply reflect the relative level of effort expended in each room.

The key attraction of the higher-taxon approach is the possibility of being able to apply it broadly across many major groups of organisms in order to estimate the distribution of wholesale biodiversity more directly. This may require more specialists, but because each has to identify only relatively few higher taxa, it should be possible to cover many more species for the same cost.

METHODS

We assessed the relationship between higher-taxon rich- ness and species richness using available data for some of the better known groups of organisms. These have been intensively and relatively evenly sampled over many years. Small, well-known groups were chosen because this is an exploratory examination of the relationship. If the generality of the relationship is established by further detailed studies, then use of higher taxa should require fewer supporting species- level data. It could then be applied to large or poorly known groups.

Family richness and species richness for resident taxa were scored among all 100 x 100 km grid squares in Britain and Ireland, using published data for ferns and their allies (Perring & Walters, 1962), and for butter- flies (Heath et al., 1984). The 100 x 100 km grid-square size for area samples is arbitrary and was adopted for convenience. Britain and Ireland are both part of a north-west European species pool and constitute only a small part of the global distribution range of most of the species involved. We also looked at the same relationship at a larger spatial scale using published data for Australian passerine birds among 5° x 5° grid squares and 10° x 10° grid squares (Blakers et al., 1984) and for North and Central American bats (Hall, 1981) among equal-area grid squares (total area of each, c. 611,000 km2; projection described in Williams, 1993). All of these data suffer from the usual limitation of being simply presence/absence records, without regard to abundance or breeding status.

For the chosen organisms, families provide a reason- able compromise between sufficiently small numbers of higher taxa to be manageable and easy to survey, and sufficiently small average range sizes for the relation- ship to be informative (i.e. not all families are repre- sented in all areas). The taxonomic treatment accepted for each group follows that used in the sources of

distribution data, except for the butterflies. In this case we follow the more recent treatment of the Nymphalidae and Satyridae as a single family (R. I. Vane-Wright, pets. comm.).

The relationships between the richness of species and higher taxa were tested using product-moment correla- tion, with a square root transformation of the data (n+0.25) to homogenise variances. An exact random model was also fitted to the data. This tests the idea that species in the faunas or floras of each area may be considered to have been drawn from the regional species pool without any effect from the species rich- ness of the higher taxa to which they belong. The upper bound of the model can exceed the number of families in the regional pool because the bounds are set at two standard deviations from the mean.

RESULTS

All of the groups show significant positive correlations (p<0.001) between the numbers of families and the numbers of species in each grid square (untransformed data are plotted in Fig. 2). In every case, numbers of families account for >79% of the variance in the numbers of species. These correlations show that it is possible to use higher taxa as a surrogate for species in surveys of richness.

For the example data sets, family richness generally falls within two standard deviations of the mean rich- ness expected from choosing species at random (Fig. 2). The worst fit of the random model to the data is found for the butterflies, which show consistently fewer families in faunas than expected. This data set is the smallest among those studied, in terms of the number of families, and so is the least representative of the aims of the higher-taxon approach. The species of two of these families have very restricted and non-overlap- ping distributions in Britain. Beyond this, for at least the fern and bat data there appears to be a slight systematic deviation, whereby grid squares with few species have fewer families than expected, and grid squares with many species have more families than ex- pected. In other words, there may be a weak tendency for species in species-poor areas to be clumped within particularly the larger families (e.g. the ferns of the Aspleniaceae and Dryopteridaceae, and the widespread 'common' bats of the Vespertilionidae). In contrast, species in species-rich areas may be slightly over- dispersed among families. This tendency is consistent with a relationship between species richness and range size for higher taxa.

DISCUSSION

The only properties required of higher taxa for estimat- ing wholesale biodiversity are (1) that their own richness distribution is predictive of the distribution of species richness, and (2) that information to map their distribu- tions is more readily acquired than for species. The first requirement is supported by our preliminary results.

214

P. H. Williams, K. J. Gaston

I

° ' °~

. . o . ° "

_

6

5

4

i

3

10

im

-@

8

•_

] 6 " m •

4

2

o

Williams, 1964; Clayton, 1972, 1974; Anderson, 1974; Bock & Farrand, 1980; Holman, 1985; Dial & Marzluff, 1989). It seems to hold both for species among higher taxa, and for other higher taxa among yet higher taxa. The distribution is likely to result in part from evolu- tionary processes (Wright, 1941; Anderson & Anderson, 1975; Glazier, 1987; Dial & Marzluff, 1989) and in part fi'om the way in which taxonomy is practised (Waiters, 1961, 1986; Clayton, 1972, 1974). So far, no model has provided a generally adequate description. Thirdly, species discrimination within many of the hyper-rich higher taxa is often very time-consuming, either because of their small body size, or because of difficulties with species concepts. Taxonomic support may be limiting because species of large body size have tended to be described before the smaller species, and the bulk of the species of the more speciose higher taxa have been described only relatively recently (Gaston, 1991b).

Considerable resources could be saved by confining taxonomic effort to higher taxa. For example, with suitably prepared samples of tropical beetles, the over- all reduction in time and cost for a specialist counting family richness rather than species richness may be of the order of 100-fold (P. M. Hammond, pers. comm.). These resources could then be re-deployed to survey a much broader range of higher taxa, in order to gain more representative estimates of overall biodiversity.

Problem of uneven sampling effort

The problems of sampling for higher-taxon richness are essentially the same as those of sampling for species richness. The most severe challenge to any attempt to measure richness at any taxonomic rank is in deciding how much effort is needed to discover the taxa that are present in an area. But with movement of individuals, species' ranges shift continuously, so there is no absolute, fixed value for local richness. Consequently any count of richness has to be measured relative to sampling effort. Either an adjustment should be made (e.g. using rarefac- tion techniques), or sampling effort should be held con- slant in some sense. For example, with enough effort, at least three families of butterflies (Pieridae, Lycaenidae and Nymphalidae) might be found in most British 100 x I00 km squares, but then so too might many more of their species. In an even more extreme example, Grinnell (1922) calculated how much time would be required to record every species in the North American bird fauna by sampling only within California.

In practice, one sampling strategy for higher taxa would be to try to effect an undirected search for in- dividual organisms. For example, certain collecting methods may be relatively 'blind' to higher taxa at some taxonomic ranks. An alternative sampling strategy would be to search in a manner directed specifically to the discovery of new higher taxa.

Problem of inappropriate rank of higher taxa

In order to preserve a close relationship with species richness for our present, pragmatic purposes, the choice of taxonomic rank at which to count higher taxa will

have to be made carefully. There has to be a compro- mise, because both the survey costs and the predictive value of the relationship should steadily decline at progressively higher taxonomic ranks. The strength of the relationship declines because an increasing propor- tion of the higher ranking taxa become widespread and uninformative, irrespective of spatial scale. Another possibility might be to combine taxa from different ranks (such as the 'family-groups' of Hammond, 1990) if this improves the relationship. The method should always be checked by comparing higher-taxon richness and species richness among at least a few local faunal and floral lists from areas at an appropriate scale.

Problem of uneven taxonomic treatment

It is not essential for predicting species richness that higher taxa are either monophyletic (natural groups including all of the descendants of a common ancestor), or that they are necessarily of comparable rank in terms of age, or of phenetic or genetic distinctiveness. This is fortunate, because higher taxa of the same nomencla- tural rank are essentially arbitrary conventions, with no objective way of comparing the rank of taxa, except in the case of sister groups (Gauthier et al., 1988).

Another separate interpretation of higher taxa is that they represent higher level units of the difference or dis- parity among organisms (Williams et al., 1991). This interpretation is more demanding, because it really requires monophyly of higher taxa. However, it does allow information on genealogical relationships to be included in diversity measures to represent the degree of disparity among the organisms (Williams, 1993). Disparity has been seen as an important component of the concept of biodiversity (Gould, 1989; May, 1990; Vane-Wright et al., 1991; Groombridge, 1992; Reid et al., 1992).

The association between higher-taxon richness and species richness could break down at the global scale, because of the differing taxonomic treatment of some groups in different continents. For example, Waiters (1961, 1986) argued that the current, non-cladistic classification of flowering-plant families has developed through a preference by authors to agglomerate any new species into the higher taxa of pre-Linnean, European folk origin. Forcing these groupings upon the flora of the rest of the world is likely to result in a global distri- bution of higher-taxon richness that is biassed towards the north temperate region by regionally uneven taxo- nomic 'splitting'. A better approach to higher rank taxonomy would be to apply names to groups within a classification inferred using cladistic methods of all organisms worldwide.

Problem of atypical regions

Most biologists will be aware of regions of the world where there are genuine differences in the taxonomic richness relationship. Hawaii is a well-known example of an archipelago with few higher taxa that is nonethe- less rich in endemic species (Simon et al., 1984; Loope et al., 1988). A difference in the richness relationship is

also likely between the terrestrial and marine environ- ments, at least for the higher taxonomic ranks, because the marine fauna includes many more phyla but fewer described species (May, 1988; McNeely et al., 1990). Further studies are needed to determine whether atypical areas present a serious challenge, or whether they merely increase scatter about a general relationship.

Application to conserving more of overall biodiversity

The key to more representative estimates of overall bio- diversity is likely to lie with indirect approaches, such as those using environmental variables, indicator groups and higher-taxon richness. N o choice between these approaches can be made until large-scale assess- ments have been carried out. However, higher-taxon richness remains a candidate for further study.

If suitable relationships between higher-taxon rich- ness and species richness were to hold for the most species-rich groups of organisms, then taken together the distribution of higher-taxon richness in these groups might provide a better first approximation to the distri- bution of a major part of overall biodiversity. The largest contributions to the overall number o f species o f macro-organisms are made by a few orders o f insects and by seed plants (Southwood, 1978; May, 1988; Barnes, 1989; Gaston, 1991a; Holloway & Stork, 1991; Hammond, 1992). F o r any effort to survey the distribu- tion o f overall biodiversity, these would be the obvious groups to look at first (Williams et al., in prep.).

ACKNOWLEDGEMENTS

Thanks to Josephine Camus, Paul Eggleton, Peter Hammond, Chris Humphries, Clive Moncrieff, Dick Vane-Wright and anonymous referees for helpful comments, and to Clive Moncrieff for supplying the exact random model in place o f our earlier simulations. The work was supported in part by N E R C research grant GR3/8335 to PHW .

REFERENCES

Adams, J.M. (1989). Species diversity and productivity of trees. Plants Today, 1989, 183-7.

Adams, J.M. & Woodward, F.I. (1989). Patterns in tree species richness as a test of the glacial extinction hypothe- sis. Nature, Lond., 339, 699-701.

Anderson, S. (1974). Patterns of faunal evolution. Q. Rev. Biol., 49, 311-32.

Anderson, S. & Anderson, C.S. (1975). Three Monte Carlo models of faunal evolution. Amer. Mus. Novit., 2563, 1-6.

Barnes, R.D. (1989). Diversity of organisms: how much do we know? Amer. Zool., 29, 1075-84.

Bibby, C.J. et al. (1992). Putting biodiversity on the map: priority areas for global conservation. ICBP, Cambridge.

Blakers, N., Davies, S.J.J.F. & Reilly, P.N. (1984). The atlas o f Australian birds. Melbourne University Press, Carlton.

Bock, W.J. & Farrand Jr, J. (1980). The number of species and genera of Recent birds: a contribution to comparative systematics. Amer. Mus. Novit., 2703, 1-29.

Brown Jr, K.S. (1991). Conservation of Neotropical environ- ments: insects as indicators. In The conservation o f insects and their habitats, ed. N.M. Collins & J.A. Thomas. Academic Press, London, pp. 349-404.

Burbidge, A.A. (1991). Cost constraints on surveys for nature

conservation. In Nature conservation: cost effective bio- logical surveys and data analysis, ed. C.R. Margules & M.P. Austin. CSIRO, Canberra, pp. 3 - 6 .

Clayton, W.D. (1972). Some aspects of the genus concept. Kew Bull., 27, 281-7.

Clayton, W.D. (1974). The logarithmic distribution of angiosperm families. Kew Bull., 29, 271-9.

Cronk, Q. (1988). Biodiversity. The key role of plants. IUCN and WWF, Gland.

Currie, D.J. (1991). Energy and large-scale patterns of animal- and plant-species richness. Amer. Nat., 137, 27-49.

Currie, D.J. & Paquin, V. (1987). Large-scale biogeographical patterns of species richness of trees. Nature, Lond., 329, 326--7.

Cutler, A. (1991). Nested faunas and extinction in fragmented habitats. Conserv. Biol., 5, 496-505.

Dial, K.P. & Marzluff, J.M. (1989). Nonrandom diversifica- tion within taxonomic assemblages. Syst. Zool., 38, 26-37.

Ehrlich, P.R. (1992). Population biology of checkerspot butterflies and the preservation of global biodiversity. Oikos, 63, 6-12.

Gaston, K.J. (1991a). The magnitude of global insect species richness. Conserv. Biol., 5, 283-96.

Gaston, K.J. (1991b). Body size and probability of descrip- tion: the beetle fauna of Britain. Ecol. Entomol., 16, 505-8.

Gauthier, J., Kluge, A.G. & Rowe, T. (1988). Amniote phylogeny and the importance of fossils. Cladistics, 4, 105-209.

Gentry, A.H. (1992). Tropical forest biodiversity: distribution patterns and their conservational significance. Oikos, 63, 19-28.

Glazier, D.S. (1987). Energetics and taxonomic patterns of species diversity. Syst. Zool., 36, 62-71.

Gould, S.J. (1989). Wonderful life. Penguin Books, London. Grinnell, J. (1922). The role of the 'accidental'. The Auk, 39,

373-80.

Groombridge, B. (ed.)(1992). Global biodiversity. Status o f the Earth's living resources. Chapman and Hall, London.

Hall, E.R. (1981). The mammals o f North America, Vol. 1, 2nd edn. Wiley, New York.

Hammond, P.M. (1990). Insect abundance and diversity in the Dumoga-Bone National Park, N. Sulawesi, with special reference to the beetle fauna of lowland rainforest in the Toraut region. In Insects and the rain forests o f South East Asia (Wallacia), ed. W.J. Knight & J.D. Holloway. Royal Entomological Society of London, London, pp. 197-254.

Hammond, P.M. (1992). Species inventory. In Global biodi- versity status o f the Earth's living resources, ed. B. Groom- bridge. Chapman and Hall, London, pp. 17-39.

Heath, J., Pollard, E. & Thomas, J. (1984). Atlas o f butterflies in Britain and Ireland. Viking, Harmondsworth.

Holloway, J.D. & Stork, N.E. (1991). The dimensions of bio- diversity: the use of invertebrates as indicators of human impact. In The biodiversity o f microorganisms and inverte- brates: its role in sustainable agriculture, ed. D.L. Hawksworth. CABI, Wallingford, pp.37-62.

Holman, E.W. (1985). Evolutionary and psychological effects in pre-evolutionary classifications. J. Classific., 2, 29-39.

Kremen, C. (1992). Butterflies as ecological and biodiversity indicators. Wings, 16, 14-17.

Landres, P.B., Verner, J. & Thomas, J.W. (1988). Ecological uses of vertebrate indicator species: a critique. Conserv. Biol., 2, 316-28.

Loope, L.L., Hamann, O. & Stone, C.P. (1988). Comparative conservation biology of oceanic archipelagos. BioSci., 38, 272-82.

MacKinnon, J. & MacKinnon, K. (1986). Review o f the protected areas system in the afrotropical realm. IUCN and UNEP, Cambridge.

May, R.M. (1988). How many species are there on Earth? Science, N. Y., 241, 1441-9,

May, R.M. (1990). Taxonomy as destiny. Nature, Lond., 347, 129-30.