Biodiversity evaluation in agricultural landscapes - Science Direct

Agriculture Ecosystems Enwronment ELSEVIER

Agriculture, Ecosystemsand Environment 62 (1997) 81-91

Biodiversity evaluation in agricultural landscapes: An approach at two different scales Peter Duelli Swiss Federal Institutefor Forest, Snow and Landscape Research, Division of Landscape Ecology, CH-8903 Birmensdorf, Switzerland

Abstract Evaluating biodiversity in agricultural areas requires two steps: (i) adequate measurement and (ii) pertinent interpretation. Since the whole spectrum of biodiversity in the sense of the Rio Convention cannot possibly be measured as such, adequate measurable correlates or surrogates have to be found. One possibility are standardized inventories to assess species diversity in specious groups of arthropods. They can provide reproducible and comparable values for estimates of site-specific biodiversity in the form of averaged rarefaction functions for the relationship between numbers of individuals and numbers of species in different habitats. Examples of 'rapid biodiversity assessment' are presented and their possibilities and limitations are discussed. At a higher integrative level, biodiversity evaluation can be based on landscape parameters. According to a proposed conceptual model, the 'mosaic concept', regional biodiversity mainly depends on structural parameters such as habitat diversity and landscape heterogeneity, and functionally on metacommunity dynamics. Approaches with a combination of both site-specific biodiversity measures and assessments of habitat diversity and heterogeneity are not yet established, but in the near future may lead to a scientifically based evaluation of the potential for increasing biodiversity by appropriate landscape management. In the light of scenarios for global change, maintaining high levels of overall biodiversity in agroecosystems may become as important for ecological sustainabilty as keeping up high abundances of presently well-adapted beneficial organisms. © 1997 Elsevier Science B.V. Keywords: Biodiversity; Evaluation; Mosaic landscape; Spatial heterogeneity; Invertebrates; Agroecosystems

1. Introduction The present paper reviews the problems of assessing local or regional biodiversity in agroecosystems. We first address the question of why biodiversity evaluation in agricultural areas requires standardized methods for qualitative and quantitative comparison. We then try to define what particular aspects of biological diversity can and should be measured in agroecosystems. The rest of the paper proposes a conceptual approach for a future link between an

evaluation of standardized biodiversity estimates at selected single sites and spatial diversity and heterogeneity in mosaic landscapes. The need for rapid and cost-saving methods is stressed, and examples from an ongoing project in Switzerland are given. 1.1. Motivation f o r evaluating biodiversity in agricultural landscapes In countries where intense agriculture over the years has led to an alarming level of ecological

0167-8809/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0167-8809(96)01143-7

82

P. Duelli / Agriculture, Ecosystems and Environment 62 (1997) 81-91

degradation, there are now more and more national efforts to revitalize agricultural landscapes and to enhance biodiversity with ecological compensation programs. 'Ecological compensation areas' have two basically different effects with regard to agricultural values: While natural or seminatural habitats such as wetlands or dry meadows are able to enrich regional biodiversity considerably, they hardly have any positive effects on beneficial arthropods. However, ecotones in the form of perennial linear landscape structures between different habitat types enhance beneficials significantly, mainly by providing habitats after harvest and for hibernation. Ecotones with high structural heterogeneity such as forest edges and hedgerows provide an enhancement for both regional biodiversity and an abundance and diversity of beneficial organisms. However, the main issue here is not the basic and general motivation for conserving or enhancing biodiversity in agriculture, but strictly the motivation for an evaluation of the present status of biological diversity in an agricultural landscape, with the intent to compare biodiversity measures either in time or in space. There are two major avenues of motivation to evaluate biodiversity in agroecosystems: (i) nature conservation and (ii) biological control against agricultural pests.

1.1.1. Species, habitat and landscape protection: the conservation aspect The motivation to evaluate conservation aspects in agricultural habitats is primarily based on ethical reasoning: What part or portion of the autochthonous biological diversity has been lost or will get lost with an intensification of agricultural production? What can be saved from extinction or brought back with extensification programs? How does sustainability relate to biodiversity? At least in densely populated and highly industrialized countries, this ethical motivation is very often combined with economical considerations (e.g. the tourist industry) or recreational benefits for local inhabitants. Biodiversity can be seen as one measure of environmental quality. Even urban people seem to have a clear preference for spending their nature-oriented holidays or recreational activities in a diverse and traditionally cultivated landscape, rather than in untouched wilderness or intensely cultivated farmland.

1.1.2. High genetic diversity to enhance the ecological resilience of agricultural habitats: the biological control aspect In times of increased traffic of agricultural products between continents and of imminant global climatic change, high species diversity and broad genetic spectra within species are considered to better maintain a sustainable balance between potentially new pest organisms and their antagonists. The World Conservation Monitoring Centre (1992) highlights the value of biodiversity in providing insurance for crop yields. Specialization in agricultural production methods may increase productivity, but at the same time it implies a decreased range of productive assets. A wide range of productive assets, however, provides a hedge against different risks relating to any one form of asset. This so-called 'portfolio effect', well known in economics, reduces the risks for losses in any particular asset, but it also reduces average productivity. 1.2. What kind of biodiversity can be evaluated in agroecosystems ? 1.2.1. Biodiversi~. in the sense of the Rio Convention: the qualitative or concept approach The term biodiversity in the sense of the Biodiversity Convention of the UN Conference on Environment and Development in Rio de Janeiro (1992) encompasses the whole range of the genetic diversity within species, the diversity of species and higher taxa, up to ecosystem diversity, and even the diversity of ecological interactions (Fig. 1). According to

Components of biological diversity

1 I. I 1.2 1.3

Diversity (quantitative aspect) Numberof species(geneticloci,highertaxa) Distributionof abundances(e.g.Shannon-lndex) Evenness

2 2.1 2.2 2.3 2.4 2.5

Biological diversity(qualitativeaspect) / 1 Geneticdiversity (within species) ~Geneticspectrmn. Species diversity / Diversityof highertaxa (family,class,phylum) 1 ~ BloOlversity Ecological interactions ( (sensu,~Rio92~) Diversityof ecosystems J

i [

Fig. 1. Quantitative and qualitative aspects of biodiversity evaluation. The 'genetic spectrum" is only one part of the term biodiversity addressed in the Rio Convention, which clearly focuses on the more complex qualitative aspects.


the Convention on Biological Diversity of 1992 (Johnson, 1993), biological diversity means 'the variability among living organisms from all sources including, inter alia, terrestrial, marine and other aquatic ecosystems and the ecological complexes of which they are part; this includes diversity within species, between species and of ecosystems'. Quite obviously, such broad 'diversity of life' cannot be measured and quantified in a comprehensive manner. And yet, for an evaluation we need measurable parameters which are as good as possible correlates or surrogates lbr intuitively 'true biodiversity'. 1.2.2. Diversity as a measurable entity: the quantitative approach The traditional scientific concept of biological diversity is based on species diversity (number of species or other taxonomic units) and on the distribution of abundances among species (Fig. 1). Species numbers and evenness among species are the bases of various mathematical indices for measuring local or regional diversities of selected taxonomic groups. While many generations of students have calculated diversity indices such as the Shannon or the Simpson index (Magurran, 1988) for plants and animals in particular biotopes, the link to biodiversity in the sense of the Rio Convention is still far from being evident or realistic. Given the wide range of aspects of biodiversity, from intraspecific genetic variability up to the complexity of trophic interactions and landscape variability, an evaluation has either to be based on conceptual parameters, encompassing regional, national, or global aspects (fight-hand side of spectrum in Fig. 2), or on site-specific empirical measurements (lefthand side of spectrum in Fig. 2). If we perceive biodiversity as one measure of environmental quality, we can tackle its evaluation in a way similar to that of climate and weather. It will require a whole set of different approaches, methods and measures. Some factors are relatively easy to assess, others are impossible to quantify in a scientific manner. Since there is no absolute measure for biodiversity, biodiversity evaluation has to be based on relative measures. Basically, there are three kinds of comparative measurements which can be attributed to the biodiversity of a specific area to be evaluated: (1) The contribution of an area to regional, na-

83

.e

Species Family

Order

Class

Phylum

Ecosystems

Complexity Fig. 2. Cumulation of the genetic spectrum and ecological interactions to visualize the term biodiversity in its non-quantifiable, conceptual sense. There is to date no unit or scale for the y-axis.

tional or world-wide biodiversity can be assessed. For instance, while mires generally have a low intrinsic biodiversity in terms of the genetic spectrum, their contribution to national biodiversity in Switzerland is very high. The reason is that today only a few mires are left in Switzerland, but the small number of highly specialized organisms living there depend on mires and thus are restricted to mires. So even though the species diversity in mires is low, each of the specialist species in a mire is an important contribution to national biodiversity. Biodiversity evaluation is a matter of scale: mires are still abundant in large parts of the northern Palearctic, where the same species which are highly threatened in Switzerland still occur in large numbers. (2) Comparison in space. Selected aspects of the biodiversity in a site, an area or region can be compared with that of neighbouring sites, areas or regions. Simultaneous comparison in space is certainly the most important kind of comparative measures in evaluating agricultural biodiversity. (3) Comparison in time. Monitoring selected aspects of biodiversity over longer time periods or before and after an impact allows an evaluation of natural succession, of environmental changes, or of specific measures taken. Regular successive measurements of biodiversity can also be the base for an assessment of sustainability.

84


1.2.3. Indicators for biodiversity With respect to the notion that the concept of biodiversity cannot be reduced to a single quantifiable entity, it is still highly desirable to have indicators or surrogates, at least for certain aspects or dimensions of biodiversity (Harper and Hawksworth, 1994). In search of indicators for monitoring biodiversity, Noss (1990) proposed using a hierarchical approach which incorporates four levels of organization: regional-landscape, community-ecosystem, population-species, and genetics. An extensive list of inventory methods and monitoring tools is offered and indicators for the three main attributes of biodiversity (composition, structure, function) are presented. So far, none of them has been used to evaluate biodiversity in agricultural landscapes. Most faunistic inventories are intended to assess some form of habitat quality or conservation values rather than biodiversity as such. Gotmark et al. (1986) in an avian case study using five conservation indices came to the conclusion that a single index for the conservation value may not be feasible and that separate indices for single evaluation criteria are a better strategy. The same is certainly true for biodiversity evaluation. Another widely proposed approach is the use of target species for conservation planning (e.g. Kremen, 1994), but the link to biodiversity is more intuitive than tested scientifically.

2. Proposal for an approach at two different scales: evaluating the mosaic pattern of an agricultural landscape and the site specific biodiversities of its mosaic patches In what follows the results and perspectives of an ongoing study in Switzerland are presented. Obviously, the proposed methods and the preliminary results are not pertinent to all agricultural regions of the world, especially not to the tropics, but in one way or the other the methods can be adapted to any agricultural situation.

2.1. Measures of site-specific biodiversity 2.1.1. Arthropod species numbers as a correlate for biodiversity In agroecosystems, we are often concerned with a particular type of crop, with a comparison between

management practices, or even with a particular field in given surroundings. Site-specific biodiversity evaluation is based on relative measures: a simultaneous comparison in space (e.g. two methods, two fields, two crops, a field and an ecological compensation area), or a comparison in time (e.g. before and after a treatment, before and after a change in agricultural politics, in financial support, or for long-term monitoring). But even in a uniform wheat field the diversity of all living organisms is too complex to be measured thoroughly. Only surrogates or correlates of true biodiversity can be quantified, and the discussion will never end as to what are the closest or best correlates to overall biodiversity. Different questions will require different methods, and an evaluation of biodiversity can have any form from a 'quick look by an expert' up to a scientific investigation lasting for several years. Usually the financial situation rather than the specific question tends to decide the effort taken for an inventory in the field. Consequently, biodiversity measures are usually based on floral inventories or bird census. Keeping in mind that more than 90% of genetic variability (e.g. in the form of species numbers) is contributed by invertebrates, it seems convincing to use them as the closest correlates to overall biodiversity. However, invertebrates are small, mobile creatures, difficult to find and even more difficult to identify. Any attempt to use them for quantifying biodiversity has to start with a standardized reduction in sample size.

2.1.2. Reducing the number of taxa The search for the best correlates to overall biodiversity among the invertebrates is influenced by the taxonomic specialists at hand and the sampling methods available. Furthermore, attractive and wellknown groups are preferred. People involved in nature conservation often specialize on a particular taxonomic group (e.g. butterflies, grasshoppers, ants, bees, syrphid flies) and try to achieve an inventory, which is as complete as possible for the area in question. By this, even very rare and faunistically interesting species may turn up, which can be of great interest for biodiversity at a national scale. However, to date and to my knowledge, there is no published record of any 'best correlating arthropod


group' in comparison with the overall biodiversity in an agricultural field or area. In order to get reproducible results for a scientific evaluation of the biodiversity at a particular site, the sampling methods have to be strictly standardized. Since the number of species found is positively correlated with the sampling effort, the effort has to be quantified exactly in any inventory (Duelli et al., 1990a). A group which has become standard in agricultural inventories of invertebrates are the carabid beetles. They combine the requirements of a standard sampling technique (pitfall trapping), are easy to preserve, mount and identify (with some exceptions), and have a positive appeal as generalist predators in crop fields. Carabids are common in most agricultural biotopes, with enough species to characterize various biotope types. But are they good correlates to overall biodiversity? Unfortunately they are not, because the number of species, and even the frequency distributions among species, are very similar in different biotope types. As soon as carabid figures are reduced to a single diversity index or

85

species number, they lose their indicative value for biodiversity. Much better in that sense are spiders and staphylinid beetles, which are also mostly predatory and get caught in the same pitfall traps along with the carabids. Pitfall traps are the best known and the most often used inventory method in agroecosystems. They have been widely used for an indication of habitat quality (Mossakowski and Paje, 1985) and for measuring nature conservation values (Margules and Usher, 1981, Eyre and Rushton, 1989). Given the wealth of empirical data already available, it seems obvious to use pitfall traps also for estimating biodiversity, although the carabids alone may not contribute much to that aspect. Taking carabids, staphylinids and spiders together, however, they might be good candidates for a valuable correlate for biodiversity.

2.1.3. Reducing sampling effort In recent years we have developed two methods, which in combination can be used to estimate and

80706050"~ 40CO

30. 20" 10. 0

i

0

I

200

I

I

400 ~ 600 480

8()0

1000 1200

I

I

I

1 4 0 0 1600 1800 2000

Individuals

Fig. 3. Average species numbers for spiders in dry meadows in Switzerland as a function of the number of individuals collected in pitfall traps. 3 is above, 2 below average, but within the standard deviation, 4 and 1 are extremely high or low diversity values. The function can be extrapolated to an estimated number of 230000 individuals per hectare (based on suction samples: Katz et al., 1989) in order to estimate the average number of spider species per hectare. Another figure to get from this kind of curve is the number of individuals to be collected (here: 480) to get 50% (here: 41) of the estimated total number of species in one hectare of a particular habitat.

86


compare site-specific biodiversity in Swiss agroecosystems: (i) Averaged standard functions for the relationship between numbers of individuals and numbers of species. Empirical data from whole season samples (usually April to September in Switzerland) of various arthropod groups, collected with pitfall traps in different biotope types, were used to calculate rarefaction curves (for an introduction to the method, see Simberloff, 1972; James and Rathbun, 1981). Fig. 3 shows an example for spiders in dry meadows in Switzerland. To produce a reliable rarefaction function, several thousand individuals have to be collected. Types and numbers of traps are of lesser importance. Per site we used five funnel traps with a diameter of 15 cm, at distances of at least 10 m to each other. Using smaller traps requires more traps per site. Compared with the often used plastic cups with perpendicular walls, funnel traps collect most arthropod groups two to three times more efficiently per cm trap diameter. Only linyphiid spiders and ants were collected in comparable numbers per cm trap diameter in both trap types (Obrist and Duelli, 1996). Fig. 3 is composed of four rarefaction curves from different dry meadows in Switzerland, sampled in different years. Any further samples (even when taken with far less effort) for biodiversity evaluation can now be compared with averaged previous experiences to obtain a crude decision whether the spider fauna in the dry meadow in question is relatively high (3: above average in Fig. 3) or even exceptionally high (4: above standard deviation in Fig. 3). Scaling and rating of diversity is easy from 1 to 4, allowing simple algebraic treatment and visualization, even if several taxonomic groups in different biotopes have to be quantified for an evaluation of regional biodiversity. We have accumulated the same type of averaged rarefaction functions for epigeal arthropods in improved grassland, maize and wheat fields, as well as for seminatural habitats such as wetlands and forest edges. A catalogue of the most important arthropod groups and habitat types in Swiss agroecosystems is in preparation. (ii) The 'minimum program 3 + 2' for estimating biodiversity in agricultural habitats. Realistically, most evaluations of biodiversity in publicly funded programs will not allow full season sampling for

financial reasons. However, any attempts to minimize the efforts have to keep track of the loss of information and the increase of scatter together with a reduction in sampling effort. A strict standardization of the methods is a prerequisite for comparable and reproducible results. With an optimization program we used the same empirical data as above to calculate which combination of two sampling periods, with a total of only five weeks finally used for further processing, give the highest yields in species numbers. For several years and for several locations in Switzerland we calculated, backwards, at what dates it would have been best to put up the traps. We found that three weeks in spring and two weeks in summer are optimal (in terms of efficiency) to collect a maximum number of species. The dates for putting up the traps may vary because of different spring weather in different years. Since arthropods are cold-blooded organisms, like plants, we can use official phenology data of the national meteorological stations to adjust for the optimum number of day degrees in different years. Experiences with Swiss meteorological conditions have shown that for a variety of trapping methods (funnel traps, yellow water pans, window interception traps) and for a number of important arthropod groups (see Table 1) the best time to start the first collecting period in spring is one week after the onset of full flowering of dandilions (Taraxacum officinale). The link with phenological data not only adjusts for the yearly variation in spring weather, but also for the altitude

Table 1 Percentage of species collected with the 'minimum program 3 + 2' (3 + 2 weeks) as compared with a full season sample of 28 weeks (100%). Identifying all specimens of the two collecting periods of 5 weeks (5 + 5 weeks) yields an average of 80% of the species Group of organisms

Collecting periods 5 + 5weeks

3 +2weeks

Carabidae Araneae Aphidophaga Staphylinidae Heteroptera Aculeata

94% 90% 76% 85% 69% 68%

79% 71% 64% 62% 55% 52%

Average

80%

64%

P. Duelli / Agriculture, Ecosystems and Environment 62 (1997) 81-91 100-

~m

90-

•~ ,

80-

E-~

-~.~ 7o• ~. 6o-

~

so_

1 trap

5

i i

i

25%

Carabidae -50%

30-

Staphylinidae

~'~

2 0-

-~s lo0

,

s~5

3+2 ....

,;2'w~k, 25

weeks

i

65

2.1.4. Reducing identification effort

2s

weeks ........

L

.....................

'

40-

~E

:

relatedness for measuring biodiversity in an evaluation of conservation priorities based on bumble bee data.

traps

i-'~. . . . . . . . . . . . . . ~............. ~ ;

weeks

~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

i

I 123 1 O0

87

I 20 0 Effort in hours

~ .............

I 300

332

40 0

Fig. 4. Percentage of species collected as a function of the effort in hours (installation of traps, weekly collection, sorting, identification at species level of all three groups) for a sampling period of 28 weeks, two optimum periods of 5 weeks, or 3 + 2 weeks (see also Table 1). The effort in hours and thus the cost for the 3 + 2 method is reduced to 20%. A further reduction from five to one trap reduces the yield in species numbers for spiders and staphylinids down to unreliable 20%.

above sea level in the range where Taraxacum officihale occurs. We suggest first collecting for a period of five weeks, emptying the traps weekly, then using only the three weeks with most individuals per week for further processing. This is to eliminate stochastic influences of exceptionally bad weather. Three weeks after the end of the first collecting period, a second collecting period of also five weeks is suggested. Of these five weeks, only the two richest samples are processed further. This reduction to process only five weeks of sampling brings down the costs to about 20% compared with a full season (seven months) sample, while still about 50% of the estimated total number of species are collected (Fig. 4). The same 'minimum program 3 + 2' used with window (interception) traps yielded between 52% and 64% of the species numbers for whole season samples of aculeate Hymenoptera, Aphidophaga (Coccinellidae, Syrphidae and Neuroptera) and Heteroptera (Table 1). Further analyses will have to show which collecting method and which taxonomic group, or which combination of groups, correlates best with the species numbers for all arthropods and plants recorded at a particular site, i.e. the best estimate we can get for site-specific biodiversity. To cover the systematic or phylogenetic spectrum rather than simply estimating species richness, Williams et al. (1991) suggested using taxonomic

Methods of 'rapid biodiversity assessment' are mainly investigated in Australia, with the suggestion to cut the costs by engaging non-specialists for identifying morphospecies as surrogates for species numbers (Disney, 1987; Cranston and Hillman, 1992; Oliver and Beattie, 1993). Focusing on plants, vertebrates or selected groups of well-known invertebrates has certainly the appeal of feasibility, but they are no proven correlates to local biodiversity. However, using a very broad spectrum of (mostly invertebrate) organisms turned out to be too tedious and too expensive (Cranston and Hillman, 1992). Oliver and Beattie (1996a) were able to show that estimates of richness of ants and spiders varied little between morphospecies and species inventories. But when comparing four types of forests, the same authors found no significant positive correlation between ant, beetle and spider species richness (Oliver and Beattie, 1996b). Another way to reduce costs would be to use higher taxonomic units such as genera or families (Gaston and Williams, 1993, Williams and Gaston, 1994). So far, all these approaches do not seem to have been tested in agricultural areas.

Eualuating the landscape mosaic pattern as surrogate for regional biodiversi~ 2.2.

a

In order to evaluate the biodiversity of an entire agricultural area or region, it would not be realistic to perform site-specific biodiversity assessments in every patch of a mosaic landscape. Ideally, of course, the different types of mosaic patches should have a known average correlative 'value' for site-specific biodiversity, e.g. in the form of standardized species numbers as in the above 'minimum program 3 + 2', or even a scaled value in comparison to known regional or national biodiversity averages established in the form of the above rarefaction functions. But as long as such measures or values are not even available for the most common agricultural habitat types, landscape ecology has to rely on surrogates rather than correlates for biodiversity evaluation. A surrogate is seen here as an intuitive estimation of biodi-

88


versity based on theories, models or concepts, while a correlate is a statistically tested indicator for biodiversity. The best known theory to explain species numbers on habitat patches is the equilibrium theory of island biogeography (MacArthur and Wilson, 1976). Originally designed for real islands, it was readily adopted in the 1970s for nature protection arguments in cultivated areas (Simberloff and Abele, 1976). To put it simply, the equilibrium theory states that biodiversity on an inland is positively correlated with the area of that island and negatively correlated with the distance to the nearest continent. Accordingly, an evaluation of biodiversity on habitat islands in an agricultural landscape would have to be based on surface area and distance to the nearest patch of the same habitat type. In agricultural areas, however, mobility or fragmentation and isolation do not seem to be the limiting factors for colonization by most species, but habitat quality. Some degree of fragmentation may in fact increase biodiversity in the long run (Quinn and Hastings, 1987; Robinson and Quinn, 1988). While the 'aereal plancton' above crop level may be very divers, only very few of those species are actually found within the vegetation or on the ground (Duelli et al., 1990b). High biodiversity in agricultural areas thus means high attractivity for potential immigrants among the bypassers and a high diversity of ecological niches to support them. 2.3. The mosaic concept f o r local and regional biodiversity in cultivated areas

Which kind of landscape offers the highest numbers of ecological niches and at the same time produces the most diverse 'aereal plancton'? At least in areas with a long history of cultivation it is not the part with 'untouched wilderness', but a mosaic landscape of traditional agriculture and forestry, mixed with patches of natural and semi-natural areas. Even suburban areas can contribute considerably to regional biodiversity. To predict and evaluate biodiversity in cultivated landscapes, the 'mosaic concept' (Duelli, 1992) offers an alternative to the theory of island biogeography. The latter has a better predictive value for distantly scattered relict islands of natural biotopes than for mosaics of cultivated land. The factors most pertinent to predict and evaluate

f

per total area L~ ;0 m

permosaicunit I exchange I

I

I

I

I

Habitat diversity

Fig. 5. Biodiversity as a function of the number of biotope types per unit area. Note that the functions shown in Figs. 5 and 6 of course are not linear in reality.

biodiversity in an agricultural mosaic landscape are (1) habitat variability (number of biotope types per unit area), (2) habitat heterogeneity (number of habitat patches and ecotone length per unit area), and (3) the surface proportions of natural (untouched), semi-natural (perennial vegetation or cultures with low input) and intensely cultivated areas (mainly annual crops and monocultural plantations). (1) Habitat variability. According to the mosaic concept, the higher the number of biotope types in an area (of at least 1 km 2, to be relevan0, the broader is the spectrum of the genetic pool in that area (Fig. 5). Every type of biotope contributes with some specialists to overall biodiversity. Even within a particular mosaic patch of one hectare, biodiversity will be higher in the more diverse surroundings, because more species from ecologically different neighbouring patches immigrate than in a more uniform area. Many organisms require different habitats for their development, feeding, overwintering and reproduction. They are more likely to find the adequate combination in a mosaic landscape than in a uniform area. (2) Habitat heterogeneity. Here the mosaic concept is most visibly opposed to the island theory (Fig. 6). According to the mosaic concept, biodiversity increases with the number (and size variation, not shown in Fig. 6) of the mosaic patches, the

P. Duelli /Agriculture, Ecosystems and Environment 62 (1997) 81-91

89

3. Conclusion Island theory

Mosaic concept t~ I

i

i

L

Habitat heterogeneity

Fig. 6. Biodiversity as a function of the number of habitat patches (with the same number of biotope types) and the length of borderlines (ecotones).

number of biotope types remaining equal. In contrast, according to island theory the number of species increases with the patch area of any particular biotope type. The explanation for the mosaic concept is on the one hand based on metacommunity dynamics: species dominances are more or less different in each mosaic patch, even of the same biotope type, because of different extinction and colonization history and different exposure to invasions by predators, parasitoids or disease organisms (Shorroks, 1991). Niche vacancies following local extinctions allow new species or ecotypes to invade and establish at least temporarily (Caswell and Cohen, 1991). On the other hand, smaller but more mosaic patches go along with more border areas. 'Soft edges' between different biotope types are known as ecotones, often harbouring a rich, specialized fauna and flora. The best known examples are forest edges, hedge rows and the shores of running and standing water. Some species profit from 'sharp edges' between different biotopes by switching between foraging in crops and hiding in trees (birds), or basking in the sun and escaping into the crop (reptiles). Many ants, bees, Coleoptera and Lepidoptera are known to switch between habitats daily for foraging, resting and reproduction. In a mosaic landscape, they all profit from spatial heterogeneity and have a good chance of finding a similar biotope type nearby.

There is an obvious need to link the two approaches at different scales, should regional biodiversity evaluation proceed from a surrogate approach to a correlation approach. While the ecology of mosaic landscapes has already become a promising field of research (Forman, 1995; Hansson et al., 1995), there is to date no established method in landscape ecology for an evaluation of biodiversity in agricultural landscapes. There is, however, a growing number of techniques to measure potential surrogates for biodiversity in landscape ecology (Forman, 1995; Turner, 1990), where spatial heterogeneity gets the highest attention (Kotliar and Wiens. 1990; Kolasa and Pickett, 1991; Turner and Gardner, 1991). The try to correlate spatial heterogeneity to species diversity in agricultural areas has so far been restricted to single taxonomic groups (e.g. Baudry and Baudry-Burel, 1982; Burel, 1989). On the basis of the mosaic concept presented here, an evaluation of regional biodiversity would have to quantify (i) the number of biotope types per unit area, (ii) the number of habitat patches per unit area, (iii) the total length of borderlines (ecotones) per unit area, and (iv) the proportions of natural, basically untouched areas, seminatural vegetation or perennial cultures, and annual crops. Additionally, for each general type of habitat patch or linear structure (ecotone), a known average 'biodiversity value' should be available on a regional or national base in order to evaluate the biodiversity of a landscape and the potential for enhancing it with ecological compensation measures. For future landscape management in heavily cultivated agricultural areas, the mosaic concept offers good possibilities to combine the necessity for keeping up the 'modem' high yields, developed in the course of the 'green revolution' since the 1960s, with the 'antiquated' and at the same time 'post-modern' pledge for more safety with respect to yield variability (sustainability) and for more regional biodiversity in general (conservation of genetic resources). Matching production and conservation biology was the leading theme of a symposium on 'Agroecology and Conservation Issues in Tropical and Temperate Regions' at Padova, Italy, in 1990 (Paoletti and Pimentel, 1992).

90

P. Duelli /Agriculture, Ecosystems and Environment 62 (1997) 81-91

Following the Rio Convention, many official organizations and research programs have started to develop standardized methods for quantifying and monitoring biodiversity at the landscape level. So far, the efforts have mainly been intentional, but hopefully in a few years all these different approaches can themselves be evaluated and coordinated.

References Baudry, J. and Baudry-Burel, F., 1982. La mesure de la diversit6 spatiale. Relations avec la diversit6 sprcifique. Utilisation dans les 6valuations d'impact. Acta Oecologica, Oecol. Applic., 3: 177-190. Burel, F., 1989. Landscape structure effects on carabid beetles' spatial patterns in Western France. Landscape Ecology, 2: 215-226. Caswell, H. and Cohen, J.E., 1991. Disturbance, interspecific interaction and diversity in metapopulations. In: M. Gilpin and I. Hanski (Editors), Metapopulation Dynamics. Academic Press, San Diego, pp. 193-218. Cranston, P.S. and Hillman, T., 1992. Rapid assessment of biodiversity using biological diversity technicians. Australian Biologist, 5: 144-154. Disney, R.H.L., 1987. The use of rapid sample surveys of insect fauna. In: M.L. Luff (Editor), The Use of Invertebrates in Site Assessment for Conservation. University of Newcastle upon Tyne, pp. 19-24. Duelli, P., 1992. Mosalkkonzept und Inseltheorie in der Kulturlandschaft. Verh. Ges. ftir Oekologie (Berlin 1991), 21: 379384. Duelli, P., Studer, M. and Katz, E., 1990a. Minimalprogramme fdr die Erhebung und Aufbereitung zooiSkologischer Daten als Fachbeitr'fige zu Planungen am Beispiel ausgew~ihlter Arthropodengruppen. Schr.-R. f. Landschaftspflege u. Natursehutz, 32:211-222. Duelli, P., Studer, M., Marchand, I. and Jakob, S., 1990b. Population movements of arthropods between natural and cultivated areas. Biol. Cons., 54: 193-207. Eyre, M.D. and Rushton, S.P., 1989. Quantification of conservation criteria using invertebrates. J. Appl. Ecology, 26: 159-171. Forman, R.T.T. (Editor), 1995. Land Mosaics: The Ecology of Landscapes and Regions. Cambridge University Press, Cambridge. Gaston, K.J. and Williams, P.H., 1993. Mapping the world's species - the higher taxon approach. Biodiv. Lett., 1: 2-8. Gotmark, F., Ahlund, M. and Eriksson, M., 1986. Are indices reliable for assessing conservation values of natural areas? An avian case study. Biol. Cons., 38(1): 55-74. Hansson, L., Fahrig, U and Merriam, G., 1995. Mosaic Landscapes and Ecological Processes. Chapman & Hall, London. Harper, J.L. and Hawksworth, D.L., 1994. Biodiversity: measure-

ment and estimation. Preface. Phil. Trans. Roy. Soc. London, B 345: 5-12. James, F.C. and Rathbun, St., 1981. Rarefaction, relative abundance and diversity of avian communities. The Auk, 98: 785-800. Johnson, S.P., 1993. The Earth Summit: The United Nations Conference on Environment and Development (UNCED). Graham and Trotman, London. Katz, E., Duelli, P. and Wiedemeier, P., 1989. Der Einfluss naturnaher Biotope auf Ph~inologie und Produktion von entomophagen Arthropoden in Intensivkulturen. Mitt. deutsch. Ges. allg. angew. Entomol., 7: 306-310. Kremen, C., 1994. Biological inventary using target taxa: a case study of the butterflies of Madagascar. Ecol. Appl., 4: 407422. Kolasa, J. and Pickett, S.T.A. (Editors), 1991. Ecological Heterogeneity. Springer-Verlag, New York. Kotliar, N.B. and Wiens, J.A., 1990. Multiple scales of patchiness and patch structure: a hierarchical framework for the study of heterogeneity. Oikos, 59: 253-260. MacArthur, R.H. and Wilson, E.O., 1976. The Theory of Island Biogeography. Princeton University Press, Princeton. Magurran, A.E., 1988. Ecological Diversity and its Measurement. Chapman & Hall, London. Margules, C. and Usher, M.B., 1981. Criteria used in assessing wildlife conservation potential: a review. Biol. Cons., 21: 79-109. Mossakowski, D. and Paje, F., 1985. Ein Bewertungsverfahren von Raumeinheiten an Hand der Carabidenbesffinde. Verh. Ges. ()kol. Bremen, 13: 747-750. Noss, R.F., 1990. Indicators for monitoring biodiversity: a hierarchical approach. Cons. Biol., 4: 355-364. Obrist, M.K. and Duelli, P., 1996. Trapping efficiency of funneland cup-traps for epigeal arthropods. Mitt. Schweiz. Ent. Ges. Oliver, I. and Beattie, A.J., 1993. A possible method for the rapid assessment of biodiversity. Cons. Biol., 7: 562-568. Oliver, I. and Beattie, A.J., 1996a. Invertebrate morphospecies as surrogates for species: a case study. Cons. Biol., 10: 99-109. Oliver, I. and Beattie, A.J., 1996b. Designing a cost-effective invertebrate survey: A test of methods for rapid assessment of biodiversity. Ecolog. Appl., 6(2): 594-607. Paoletti, M.G. and Pimentel, D., 1992. Biotic Diversity in Agroecosystems. Elsevier, Amsterdam. Quinn, J.F. and Hastings, A., 1987. Extinction in subdivided habitats. Cons. Biol., 1(3): 198-208. Robinson, G.R. and Quinn, J.F., 1988. Extinction, turnover and species diversity in an experimentally fragmented California annual grassland. Oecologia, 76: 71-82. Shorroks, B., 1991. Competition on a divided and ephemeral resource: a cage experiment. Biol. J. Linnean Soc.. 4 3 : 2 1 1 220. Simberloff, D., 1972. Properties of the rarefaction diversity measurement. Am. Naturalist, 106: 414-418. Simberloff, D.S. and Abele, L.G., 1976. Island biography theory and conservation practice. Science, 191: 285-286. Turner, M.G., 1990. Spatial and temporal analysis of landscape patterns. Landscape Ecol., 4: 21-30.

P. Duelli / Agriculture, Ecosystems and Em, ironment 62 (1997) 81-91 Turner, M.G. and Gardner, R.H. (Editors), 1991. Quantitative Methods in Landscape Ecology: The Analysis and Interpretation of Landscape Heterogeneity. Springer-Verlag, New York. Williams, P.H. and Gaston, K.J., 1994. Measuring more of biodiversity: can higher taxon richness predict wholesale species richness? Biol. Cons., 67:211-217.

91

Williams, P.H., Humphries, C.J. and Vane-Wright, R.I., 1991. Measuring biodiversity: taxonomic relatedness for conservation priorities. Aust. Syst. Bot., 4: 665-679. World Conservation Monitoring Centre, 1992. Global Biodiversity: Status of the Earth's Living Resources. Chapman & Hall. London.