Chapter 29 The economics of biodiversity

The conservation of biodiversity is a major environmental issue, one that promises to remain at or near the top of the environmental agenda for the foreseeable future. The conservation of biodiversity is a major environmental issue, one that promises to remain at or near the to

THE ECONOMICS OF BIODIVERSITY

3.1 Use value and existence values of individual species 1525

5 Incentives to conserve and conservation policy 1544

Handbook of Environmental Economics, Volume 3 Edited by K.-G Mäler and J.R Vincent

© 2005 Elsevier B.V All rights reserved

DOI: 10.1016/S1574-0099(05)03029-9

The conservation of biodiversity is a major environmental issue, one that promises toremain at or near the top of the environmental agenda for the foreseeable future Theloss of biodiversity affects human welfare as well as being lamentable for its own sake.Humans depend on natural systems to produce a wide variety of ecosystem goods andservices, ranging from direct use of certain species for food or medicines to ecosys-tem functions that provide water purification, nutrient retention or climate regulation.Threats to biodiversity include habitat loss and fragmentation, the introduction of non-indigenous species, over-harvesting, pollution, changes in geochemical cycles and cli-mate change Sustaining biodiversity in the face of increasing human populations andincreased human economic activity promises to be a major challenge Economists have

an important role to play in helping to develop and evaluate conservation strategies.Because biodiversity is at risk in large part because of human activity, finding ways

to conserve biodiversity will come from better understanding and management of man affairs, not from better biology alone Economists can help set priorities to allocatescarce conservation resources where they will do the most good Economists can helpdesign incentive schemes to make conservation policy both effective and efficient Eco-nomic methods can shed light on what are the most valuable components of biodiversity,including analysis of species existence value, the value of bioprospecting and the value

1 Introduction

The second half of the 20th century was, in many respects, good to Homo sapiens.

Human population more than doubled between 1950 and 2000, growing from mately 2.5 billion in 1950 to just over 6 billion in 2000 (U.S Bureau of Census) Humanhealth, nutrition and average life expectancy improved dramatically [Johnson (2000)].The value of economic activity increased by over 400% over the second half of the 20thcentury [Delong (2003))] The same period was not so good to many of the other species

approxi-on the planet Some ecologists fear that we may now be witnessing the sixth great tinction wave on the planet Though evidence is fragmentary, current rates of speciesextinction are estimated to be several orders of magnitude above background or naturalextinction rates [NRC (1995), Lawton and May (1995), Pimm et al (1995)] The area

ex-of natural habitat has declined as humans have converted lands to agriculture, managedforests or urban development Roughly half of useable terrestrial land (i.e., land that isnot tundra, ice, boreal, rock, desert) is devoted to grazing livestock or growing crops[Tilman et al (2001)] There is roughly half the area of forest now than existed whenagriculture began 8000 years ago [WRI (1998)] Increased movement of goods has led

to introductions of nonindigenous species with occasional substantial consequences fornative ecosystems Over-harvesting of fish and game species, climate change, pollu-tion, and changes in geochemical cycling has also threatened many species Given thisexperience, ecology may supplant economics as “the dismal science”

Conserving biodiversity has become a major environmental issue In the words ofecologist Simon Levin: “The central environmental challenge of our time is embodied

in the staggering losses, both recent and projected of biological diversity at all levels,from the smallest organisms to charismatic large animals and towering trees.” [Levin(1999, p 1)] The loss of biodiversity may be lamentable for its own sake but it alsohas impacts on human welfare Humans depend on natural systems to produce a widevariety of ecosystem goods and services, ranging from direct use of certain speciesfor food or medicines to ecosystem functions that provide water purification, nutrientretention or climate regulation Cultural and spiritual values are also tied to elements ofbiodiversity

The great concern for conserving biodiversity, in contrast to concerns for specificendangered species or particular ecosystems, is a relatively recent phenomenon Theformation of the Society for Conservation Biology in 1985, the beginning of the jour-

nal Conservation Biology in 1987, and the publication of the edited volume versity [Wilson (1988)] serve as useful benchmarks signaling the beginning of broad

BioDi-scientific and policy interests in the conservation of biodiversity By now there arethousands of journal articles and books devoted to various aspects of conservation (apartial biodiversity bibliography containing approximately 5000 entries can be found athttp://www.apec.umn.edu/faculty/spolasky/Biobib.html)

While biological scientists have a central role in researching biodiversity, economistshave begun to play an important and expanding role [a collection of recent biodiversityarticles by economists is contained inPolasky (2002)] Biodiversity is at risk largely

because of human activity Therefore, conservation solutions will come from better derstanding and management of human affairs, not from better biology alone Sincethere are limited budgets for conservation that cannot support all worthy conservationprojects, economists can help set priorities to allocate scarce resources where they will

un-do the most good Economists can help design incentive schemes to make conservationpolicy both effective and efficient Economic methods can shed light on what are themost valuable components of biodiversity

In this chapter we review the recent economics literature on biodiversity We begin inSection2by discussing various ways to define and measure biodiversity In Section3wediscuss the sources of value generated by biodiversity and various empirical measures

of value Section4covers strategies to conserve biodiversity in light of the main threats,namely habitat loss and invasive species Section5discusses incentives (and disincen-tives) for conservation as well as policies whose goal is to conserve biodiversity Weoffer brief concluding comments in Section6

2 Measures of biodiversity

The term biodiversity has been defined in a number of different ways Most of the sures of diversity developed and used by economists are defined as measures of thejoint dissimilarity among a set of species There is, however, another strand of eco-logical literature that defines diversity in terms of the relative abundances of specieswithin a community We begin this section with a review of measures based on relativeabundance and then review measures based on joint dissimilarity

mea-2.1 Measures based on relative abundance

Perhaps the most common way in which the word diversity has been used in ecology

is to characterize the relative abundances of species within a community [e.g., May(1972), Magurran (1988)] The relative abundance of a species in a community is de-fined as the proportion of individual organisms in the community that belong to that

species Consider a community containing s species and let π = (π1π2 π s ) be the vector of relative abundances with for all j and
s

j=1π j = 1 In qualitative terms, the

community is said to be diverse if all of the elements of π are close to 1/s There is

some evidence that communities that have recently been subjected to disturbance havelow diversity A possible explanation is that, following disturbance, recolonization is led

by opportunistic species with high fecundity (so-called r-strategists) whose numbers

quickly dominate the community Over time, less fecund, but longer-lived species

(so-called K-strategists) are able to compete successfully, thereby making the community

more diverse The extent to which diversity reflects dynamical properties like stabilityand resilience remains an open question The chapter on ecosystem dynamics bySimonLevin and Stephen Pacala (2003)in Volume 1 of this handbook discusses some aspects

of this question and other matters related to the measurement of biodiversity

This qualitative notion of diversity is not amenable to analysis and, during the 1970s,there was a burst of attention to devising quantitative measures that capture this notion.

An important contribution byPatil and Tailie (1977)was to define the diversity of a

community with relative abundance vector π by average rarity:

where r j is a measure of the rarity of species j Different choices of rarity measure yield

different diversity measures.Patil and Tailie (1977)focused on the one-parameter

which is called the Simpson index and gives the probability that two individuals sampled

at random are of different species; and the limiting form

It is fair to say that the notion that a high level of diversity of this type is preferable

to a low level is without a clear basis in either ecology or economics.Weitzman (2000)attempts to provide such a basis through a model of the relationship between the abun-dance of a crop and the number of pests or pathogens specific to that crop Briefly, underthis model:

S j = kB z

j ,

where S j is the number of pests specific to crop j and B jis the total biomass of the crop.Weitzman (2000)imagines that this biomass is divided into B j separate patches each

of unit biomass If each pathogen has a probability ε of destroying a patch of biomass,

then (under two independence assumptions) the probability of complete extinction of

the crop is:

P j (B j )=1− (1 − ε) kB zBj

.

Weitzman (2000)uses the approximation that, for small ε, P j (B j ) ∼ = (kB z

j ε) B j to showthat the set of relative abundances that minimizes the probability of a loss of all crops

maximizes the diversity measure D0(π ) The essential tension in this optimization lem is between the safety provided by a large number of patches and the correspondinglarge number of potentially harmful pests

prob-2.2 Measures based on joint dissimilarity

The focus of the rest of this section is on quantitative measures of diversity that areintended to reflect the joint dissimilarity of a collection of species One practical moti-vation for work in this area has been the need to evaluate policies aimed at protectingspecies from extinction It is worth stressing that, as the goal of such policies is to pre-serve species from extinction, this kind of diversity measure should be sensitive only toextinctions (or changes in extinction probabilities) and not to ecological changes (e.g.,

in population size or abundance distributions) This is not to say that maintaining lations is unimportant – indeed, it is a critical policy instrument in preventing extinction– only that it is not the ultimate goal In cases where species are harvested for food orsport, increasing the population size will have utility in and of itself There is a largebioeconomics literature that analyze these issues [Clark (1990)] and we will not con-sider population size issues further in this section

popu-The standard measure by which policies aimed at preventing extinction are ated is the number of species that they protect However, this measure of diversity –species number – takes no account of the relative differences between species A pol-icy that protects a large number of species covering a small number of genera may

evalu-be, in some sense, inferior to a policy that protects a smaller number of species ing a larger number of genera This point was made byVane-Wright, Humphries andWilliams (1991), who went on to propose measures of the diversity of a collection ofspecies based on the phylogenetic tree connecting them Under the simplest of thesemeasures, each species in the tree is assigned a numerical value inversely proportional

cover-to the number of nodes in the tree associated with it For a given species, this valuereflects the number of closely related species The diversity of a collection of species isthen found by summing these values for the species in the collection By measuring thediversity of a collection of species by combining values assigned to the species in thecollection, this measure cannot distinguish between the case in which a pair of closelyrelated species are both in the collection and the case in which only one is This cre-ates a problem: it would not make sense to depreciate the value of preserving a specieswith many close relatives when these close relatives are doomed to extinction Otherproposed measures suffer from a similar problem [e.g.,Haney and Eiswerth (1992)]

To formalize the measurement problem, consider a collection S = (s1, s2, , s n )

of n species or other types and let d(s , s ) be the distance or dissimilarity between

species s j and s k (neither of which needs be in S) The problem is to define a negative real-valued function D(S) that measures the diversity of S It is natural to require D to satisfy the following conditions [Weitzman (1992)] First, diversity should not be reduced by the addition of a species to s That is, if S and Sare two collections

non-of species with S ⊂ S, then D(S) < D(S) Second, diversity should not be increased

by the addition of a species that is identical (in the sense of having 0 dissimilarity) to

a species already in S That is, if s◦ is a species not in S, then D(S ∪ s◦) = D(S)

if and only if d(s◦, s i ) = 0 for some si ⊂ S Third, diversity should not decrease

with an unambiguous increase in the dissimilarities between species That is, for a

one-to-one mapping of S onto S such that d(s i , s j )  d(s

D W (S)= maxD W (S − si ) + d(si , S − si )

, where S −si denotes the collection S with species s i removed; the distance d(s i , S −si ) between species s i and the collection S − si is defined as the minimum of the distances

between s i and the species in S − si ; and the maximum is taken over the species s i in S.

In the important case where the distances between species are ultrametric (so that, forany set of three species, the largest two distances between species are equal), the speciescan be represented by a planar tree, and then Weitzman’s measure corresponds to thetotal length of the tree This is arguably the only sensible measure of pure diversity inthis case Other diversity measures have been proposed [e.g.,Crozier and Kusmierski(1996), Faith (2002)], but these tend to be ad hoc

It is worth noting at this point that Weitzman’s measure and other early measures ofdiversity were not directly connected to any theory of economic (or ecological) value.This is not to say that the qualitative connection had not been made Ecologists havelong believed that diversity in nature supports the stability and resilience of ecosystems[despite some surprising suggestions to the contrary,May (1972)] Diversity, therefore,has economic value arising not only from direct benefits but also from indirect benefitsfrom ecosystem functions that generate valuable ecosystem services, which we discuss

in more detail in Section3.3 Economists have developed the argument that, to theextent that similar species provide similar benefits and suffer similar susceptibilities, it issensible to maintain a diverse portfolio of species The first attempt to base a measure ofdiversity on a theory of value was byPolasky, Solow and Broadus (1993) Using a highlystylized probability model of substitutability of species in providing a single benefit,Polasky, Solow and Broadus (1993)derived a measure of the diversity of a collection ofspecies that reflected the probability that at least one species in the collection providedthe benefit The measure satisfied the three requirements listed above

The diversity measure ofPolasky, Solow and Broadus (1993)was based on a plete (if stylized) model of substitutability In later work,Solow and Polasky (1994)took a slightly different approach Suppose that interest in species conservation arises

com-from the possibility of species providing a benefit, such as a cure for a disease, in thefuture Suppose further that having more than one species that provides this benefit is

no better than having a single species that provides it Let B i be the event that species s i

provides this benefit The event that the collection S also provides this benefit is given

The expected benefit of S is p(S)V , where p(S) = prob(B(S)) and V is the fixed unit

value of the benefit Because V is fixed, p(S) provides a basis for comparing different

collections of species

In the absence of specific information,Solow and Polasky (1994)assumed that the

probability of the event B i that species s i would provide the benefit is an unknown

constant p that does not depend on i They also assumed that the conditional probability

of B i given the event B j that species s jprovides the benefit is:

prob(B i | Bj ) = p + (1 − p)fd(s i , s j )

, where f is a known function satisfying f (0) = 1; f (∞) = 0; and f  0 Although it

is not possible to obtain the n-variate probability p(S) from the univariate probability p and the set of conditional probabilities prob(B i | Bj ),Solow and Polasky (1994)used aprobability inequality due toGallot (1966)to show that a lower bound on p(S) is given

by p2D SP (S), where

D SP (S)= et F−1e

for the n-by-n matrix F = [f (d(si , s j ))] In summary, a lower bound on the expected

value of a collection S is an increasing function of the quantity D SP (S).Solow and lasky (1994)went on to show that, under reasonable assumptions about the function f ,

Po-this quantity also meets the three main requirements of a diversity measure Moreover,

D SP (S) is bounded below by 1 and above by n, so it has the interpretation as the tive number of species in S.

effec-The main disadvantage of D SP (S) as a diversity measure is that it requires the cation of the function f that, in some sense, measures the “correlation” between species

specifi-as a function of the dissimilarity between them On the other hand, from a practical

per-spective, it may be a disadvantage of D W (S) that it can increase without bound with

dissimilarity A more serious objection to the practical use of these and other diversitymeasures in actual conservation decision-making is that their information requirementsare utterly unrealistic Except in extremely unusual situations, conservation decisionsinvolve large numbers of species from a wide variety of taxonomic groups whose iden-tities – let alone genetic dissimilarities – are simply unknown The numbers of speciesthemselves are also unknown and comparisons based on estimated species number arefraught with problems arising from sampling An attractive option in this situation is toaim conservation efforts at conserving a diverse collection of habitats, on the assump-tion that dissimilar habitats tend to support dissimilar species Of course, this merely

transforms the problem to one of assessing the diversity of a collection of habitats Inthis case, however, the information needed to construct dissimilarities between habitats(e.g., topography, climate, etc.) may be more readily available Note that, in contrast

to the species case, there is no reason to assume that habitats are related through ananalogue to a phylogenetic tree For this reason, diversity measures based on such astructure have no particular appeal

3 Sources of value from biodiversity

Some issues related to the value of biodiversity were touched upon in previous tions, for example the utilitarian motivation for the diversity measure ofSolow andPolasky (1994) For the most part, however, discussion of diversity measures and value

sec-of biodiversity have been conducted in largely separate literatures In this section wereview economics literature on the value of biodiversity (For background on valuation

methods, see Volume 2 of this handbook, titled Valuing Environmental Changes.)

Bio-diversity is a broad term encompassing everything from genes to species to ecosystems.Value from biodiversity can arise at any of these levels We begin with value generated

at the species level

3.1 Use value and existence values of individual species

Humans have recognized the direct use value of other species, at least implicitly, for

as long as our species has existed For millennia, humans depended upon successfullyhunting animal species and gathering plants species The switch to domesticated agri-culture changed the form but not the substance of our reliance on other species Many ofthe direct use values of species which humanity relies upon for food and fiber are welldocumented by agricultural economists, fishery economists and forestry economists.Even for species that are not grown or harvested commercially, there is a long traditionwithin economics of estimating the value of recreational hunting and fishing [see, forexample,Walsh, Johnson and McKean (1988), Markowski et al (1997), Rosenbergerand Loomis (2001)]

The increased interest in the conservation of biodiversity, especially conserving dangered species, brought about a new type of species valuation effort focused onspecies existence value rather than on direct use value Beginning in the 1980s, econo-mists began to use contingent valuation surveys to ask people what they were willing-to-pay to conserve particular rare or endangered species In reviewing studies covering

en-18 different species,Loomis and White (1996, p 249)concluded “that the contingentvaluation method can provide meaningful estimates of the anthropocentric benefits ofpreserving rare and endangered species.” Estimates of annual willingness-to-pay variedfrom a low of $6 per household for the striped shiner to a high of $95 per householdfor the spotted owl The estimates of willingness-to-pay tended to be higher for more

charismatic species and for situations with greater increases in population sizes, as onewould expect.

Not all economists agree that the contingent valuation method can “provide ful estimates” of value for conserving species.Brown and Shogren (1998)note that byaggregating the estimates reported byLoomis and White (1996)over all households, theimplied willingness-to-pay to protect less than 2% of endangered species exceeds 1%

meaning-of GDP, which they remark seems “suspiciously high” Responses to contingent ation surveys on conserving species may exhibit “embedding effects” [Kahneman andKnetsch (1992)] A survey of willingness-to-pay for protecting a collection of speciesmight generate estimates similar to the willingness-to-pay for protecting an individualspecies.Desvouges et al (1993)found similar estimates for willingness-to-pay for pre-venting 2000, 20,000 and 200,000 bird deaths Survey responses may reflect the value

valu-of protecting an ecosystem (e.g., the value valu-of old growth forests rather than the value valu-ofspotted owls), the environmental more generally, or the “warm-glow” of contributing to

a worthy cause [Andreoni (1989, 1990)]

On a different tack,Stevens et al (1991) found that many people object to surveyquestions that try to elicit a monetary value for species existence They found that thevast majority of survey respondents thought conserving species was important but theyrefused to state that they would pay a positive amount for conservation.Stevens et al.(1991, p 268)attribute the unwillingness to state a willingness-to-pay as arising frommoral or ethical concerns about asking “people to choose between ordinary goods (in-come) and a moral principle.”

3.2 Biological prospecting

With arguments about existence values unlikely to be viewed as conclusive, vationists looked for other means to show that conserving biodiversity would makefinancial, as well as moral, sense One argument used extensively in the early 1990swas that conserving species preserved option value: species might contain valuablecompounds that would yield valuable pharmaceuticals or other products at some fu-ture date [see, for example,Wilson (1992)] If the species were to go extinct this optionvalue would be lost Determining the magnitude of this option value was the centralfocus of a literature on bioprospecting

conser-The early bioprospecting literature produced a wide range of estimates of the value ofconserving a species for pharmaceutical purposes, from $44 [Aylward (1993)] to $23.7million [Principe (1989)] per untested species The value of conserving an untestedspecies was derived by multiplying the probability of successfully identifying a com-mercially valuable product with the average revenue generated by a successful product.Simpson, Sedjo and Reid (1996)criticized this method This simple procedure generatesestimates of the average value rather than the marginal value of an untested species Be-cause multiple species may contain the same compound and in this sense be redundant,marginal values are likely to be far less than average values

Simpson, Sedjo and Reid (1996)developed a model of sequential search that takes

species redundancy into account In their model each of N species has an identical probability p of containing a success Testing each species costs c In the event of a success, revenue R is obtained and the search is terminated Under this model, it is shown that the value of a collection of N species is

v(N ) = V (N + 1) − V (N) = (pR − c)(1 − p) N

The value of the final species is the expected profit of testing the species, pR − c,

multiplied by the probability that tests on all prior species tested have been unsuccessful,

(1 − p) N If p is small, the expected profit of testing a species is likely to be small, leading to low marginal value On the other hand, if p is large, the species are likely to be

redundant for bioprospecting purposes, leading to low marginal value Using evidence

to assign reasonable parameter values for revenues and costs, the number of speciesand the expected number of potential products,Simpson, Sedjo and Reid (1996)solvefor the probability that generates the maximum marginal species value For flowering

plants (with an N of 250,000), they find that when p = 0.000012, the marginal value

of a species may be as high as $9431

Simpson, Sedjo and Reid (1996)then use this maximum estimate for value of ginal species along with species–area curves and estimates of endemism (species unique

mar-to an area) mar-to estimate the maximum marginal value of a hectare of land in each of

18 global biodiversity hotspots These estimates range from a maximum of $20.63per hectare in Western Ecuador to only $0.20 per hectare in the California FloristicProvince On the basis of their theoretical and empirical results,Simpson, Sedjo andReid (1996)conclude that the incentive to conserve biodiversity for bioprospecting pur-poses is almost certainly too small to offset the opportunity cost of development.Polasky and Solow (1995)used a similar model to value a collection of species If the

probability of success on any given trial is p and the revenue upon success is R, then the expected value of a collection of N species is V (N ) = R[1 − (1 − p) N], which is

the same as bySimpson, Sedjo and Reid (1996)when c= 0.Polasky and Solow (1995)considered two variants of the simple model to allow for imperfect substitutes amongspecies that generate success for the same product, and dependence in probabilities ofsuccess across species that relate to genetic similarity Both extensions are motivated

by the experience of bioprospecting When taxol was found in the bark of the Pacificyew tree, there was an intensified search of related species It was found that the needles

of the European yew tree could be used to get taxotare, an imperfect substitute fortaxol With imperfect substitutes, the marginal value of species need not fall as fast asindicated bySimpson, Sedjo and Reid (1996) On the other hand, accounting for speciesinterrelationships tends to reduce the marginal value of species

Rausser and Small (2000) challenge the empirical conclusions of low value frombioprospecting found in Simpson, Sedjo and Reid (1996) The existence of prior in-formation makes it unlikely that all species will have the same probability of success

in yielding a valuable product Under the assumption that the probability of success isindependent and differs across species, it is optimal to organize the search in order ofdescending success probability of success When this is done, the value of conserving aspecies with a high probability of success may be large.Rausser and Small (2000)applytheir model to the empirical case examined bySimpson, Sedjo and Reid (1996), withthe assumption that probabilities are proportional to the density of endemic species ineach region, which range from one in ten thousand (Western Ecuador) to one in a mil-lion (California Floristic Province) Rausser and Small find optimal search yields anincremental value of $9177 for the most promising hectare of land in Western Ecuadorcompared with a marginal value of only $20.63 for the same hectare inSimpson, Sedjoand Reid (1996) This result suggests that the benefits of protecting biodiversity hotspotsfor future biological prospecting may indeed outweigh the costs

Costello and Ward (2003) show, however, that the difference in results betweenRausser and Small (2000) and Simpson, Sedjo and Reid (1996) does not come fromwhether search is optimally ordered or random, but rather from an assumption of differ-ent parameter values In fact, the value of a hectare in Western Ecuador is $9177 whenconducting an optimal ordered search and only drops to $8840 with random search

It appears as though the answer to the bioprospecting question may be dependent, and will likely hinge critically on the quality of information available tothe bioprospector, for example indigenous knowledge about what species are likelyprospects, and the opportunity cost of land Neither of these two features has been care-fully analyzed empirically in this literature We will return to the issue of bioprospecting

context-in Section5, where we consider its impacts on incentives for conservation

3.3 Biodiversity and ecosystem services

In the previous sections, it was the elements of biodiversity, the genes or the species, thatwere the focus of analysis as being the sources of value An alternative line of reasoningfocuses on ecosystems as a whole, rather than the individual parts (genes or species),

as being the primary sources of value Ecosystems provide a wide range of goods andservices of potential value to people For example, a wetland may provide flood control,absorbing high waters and gradually releasing water over time It may also filter andretain nutrients and pollutants thereby providing cleaner water downstream Ecosystemservices include provision of clean air and water, climate regulation, mitigation of nat-ural disturbances, waste decomposition, maintenance of soil fertility, pollination, andpest control, among other things [Daily (1997)]

In principle, quantifying and valuing ecosystem services is no different from tifying and valuing human produced goods or services In practice, however, valuingecosystem services is problematic To date there have been few reliable estimates of thevalue for most ecosystem services like those mentioned in the previous paragraph There

quan-are several difficulties in estimating the value of ecosystem services First, the currentstate of ecological knowledge may be insufficient to link ecosystem condition and func-tioning to the provision of ecosystem services In other words, we may not understandecosystem “production functions” well enough to quantify how much service is pro-duced, or how changes in ecosystem condition or function will translate into changes

in amounts of ecosystem services produced [Daily et al (2000)] Second, even if tification of the services is possible, the current state of economic methods may not besufficient to yield reliable estimates of the value of services For some ecosystem ser-vices, such as flood control benefits, establishing reasonable estimates of value may notpresent great difficulties For other services it may be exceedingly difficult to assign avalue, such as the value of habitat to lower the probability of species extinctions Finally,valuing ecosystem services requires integrating ecological knowledge with economics,which requires cooperation between ecologists and economists Such cooperation hasbegun [e.g.,Brown and Roughgarden (1995), Perrings and Walker (1995), Carpenter,Ludwig and Brock (1999), Tilman et al (2001)] but is still more the exception than therule

quan-Some of the most successful efforts to estimate the values of ecosystem serviceshave focused on the production of specific tangible outputs, such as the production offish and game species Such outputs tend to be readily measurable and may even havemarket prices (e.g., commercially harvested fish) The ecosystem (habitat) is a necessaryinput into the production of the output (species) Focusing on the input side, such worknaturally fits into a classification of ecosystem services One can just as easily focus onthe output side, however, in which case such studies are best considered as studies ofthe values of individual species (as covered in Section3.1) Some research in economicshas focused on how changes in ecosystem conditions translate into changes in the value

of output [e.g.,Hammack and Brown (1974), Lynne and Prochaska (1981), Ellis andFisher (1987) and Swallow (1990)].Barbier (2000)provides a review of a number ofpapers that estimate the change in the value of fisheries production with changes incoastal wetlands and mangrove ecosystems

One widely cited example of valuing an ecosystem service is the provision of cleandrinking water for New York City New York City gets a sizeable fraction of its waterfrom watersheds in the Catskills Increased housing development with septic systems,runoff from roads, and agriculture were causing water quality to decline Continueddeclines in water quality would have forced the U.S Environmental Protection Agency

to require New York City to build a water filtration plant The total present value cost

of building and operating the water filtration plant was estimated to be roughly $6 to

$8 billion [Chichilnisky and Heal (1998)] Instead, New York City decided to invest

$1 to 1.5 billion to conserve the Catskills watersheds to avoid building the water tion plant Preserving the ecosystem was a far cheaper way to provide clean drinkingwater The value of the ecosystem service here is the savings provided by avoided cost.Avoided cost is not the value of clean water Rather it is the cost of replacing the ecosys-tem service with some human engineered alternative Measures of avoided cost can beused as measures of the value of ecosystem services, but only under certain conditions,

filtra-namely that an alternative human engineered alternative exists and that the cost of such

an alternative does not exceed the value of the benefit provided by the service Bothconditions hold for the New York City example

At the other end of the spectrum are controversial efforts that attempt broad scale

or even global valuation of ecosystem services.Costanza et al (1997)estimated thatthe mean value of global ecosystem services was $33 trillion annually, which is greaterthan the value of global GDP at the time ($18 trillion) The paper garnered a lot ofattention and it remains probably the most widely known work on valuing ecosystemservices, especially among non-economists, despite being roundly criticized by manyeconomists [Bockstael et al (2000), Toman (1998)] If what is being valued is the life-support system provided by the Earth’s ecosystems, then as Michael Toman noted, theestimate of $33 trillion is “a serious underestimate of infinity”

Other papers have since stressed the importance of more focused analysis thatmatches the scale of analysis for ecosystem valuation to the scale of management ques-tions [Daily et al (2000), Balmford et al (2002)] For example,Balmford et al (2002)used studies of specific ecosystems to ask whether conservation or development optionsgenerate greater value Their study shows that conservation options are preferred, often

by wide margins Earlier work byPeters, Gentry and Mendelsohn (1989)reached a ilar conclusion on conserving tropical forests However, much of the work to date onecosystem services requires making large leaps to overcome lack of data or understand-ing Much greater understanding of ecosystem functioning, how these functions changewith management actions, and how such changes impact on human values is requiredbefore firm conclusions can be reached

sim-One question that has received considerable attention in ecology over the past decade

is whether systems with greater diversity, in the sense of having more species, arealso more productive Results from experiments in grassland plots [Tilman, Wedin andKnops (1996)] and in controlled environments [Naeem et al (1994, 1995)] found thatincreasing the number of species in the system tended to increase system productivity.Similar results were reported byHector et al (1999)in experiments across a number ofEuropean countries

Several reasons have been put forward to explain the diversity–productivity link[Tilman, Lehman and Thomson (1997), Tilman (1999)] When species differ in theirproductivity, a collection with more species is more likely to include high productiv-ity species in the mix (the sampling effect) In heterogeneous environments, havingmore species will generally allow the collection of species to better utilize all ecologi-cal niches and so be more productive (the niche differentiation effect).Tilman, Lehmanand Thomson (1997)develop a model of niche differentiation that is formally similar

to thePolasky, Solow and Broadus (1993)model of the probability that a set of specieswill contain a given trait Both models show how coverage increases, either in nichespace or genetic space, when more species are added The model can also be interpreted

as showing that diversity allows greater productivity on average when environmentalconditions vary over time

Greater diversity has also been linked to lower variance of ecosystem productivity[Tilman and Downing (1994), McGrady-Steed, Harris and Morin (1997), Naeem and Li(1997)].Tilman, Lehman and Bristow (1998)explain this result using reasoning fromeconomics, which they call the portfolio effect Increasing the number of species thatfluctuate independently will decrease system volatility, just as increasing the number ofindependent assets in a financial portfolio will decrease the volatility of returns Greaterdiversity leads to diversification, which leads to lower variance.

In a closely related line of reasoning,Perrings et al (1995, p 4)state that “The portance of biodiversity is argued to lie in its role in preserving ecosystem resilience,

im-by underwriting the provision of key ecosystem functions under a range of tal conditions.” Conserving biodiversity maintains species that may look unimportantfor ecosystem function under current conditions, but who may play a crucial role in

environmen-a drought, pest infestenvironmen-ation or other shock [Wenvironmen-alker, Kinzig environmen-and Lenvironmen-angridge (1999)] Asabove, conserving diversity can lead to a greater probability of having a productive mix

of species under a range of potential environmental conditions Resilience has been fined in two main ways in the ecological literature First, resilience can be defined interms of how quickly an ecosystem returns to equilibrium after a shock Second, re-silience can be defined in terms of the magnitude of shock that can be absorbed withoutthe ecosystem flipping into the basin of attraction of another equilibrium state Undereither definition, if the equilibrium is desirable, then greater resilience will tend to in-crease welfare, which is what most of the literature on biodiversity and resilience hasimplicitly assumed However, increased resilience in an ecosystem in an undesirableequilibrium will not in general be beneficial

de-Some ecologists have been skeptical of the findings linking diversity to greater ductivity, stability, or resilience [e.g., Grime (1997), Wardle et al (1997)] Huston(1994)notes that there is an inverse relationship between plant diversity and ecosystemproductivity across ecosystems Examples of low-diversity, high-productivity systemsinclude salt marshes, redwood, bamboo and Douglas fir forests Critics of the “pots andplots” experiments point out that the processes operating under artificial selection in di-versity experiments are quite different from processes operating under natural selection

pro-in real ecosystems Therefore, experimental results may not be very pro-informative aboutthe role of diversity in ecosystem functioning

If the experimental results are correct, this suggests that the loss of biodiversity willlead to the loss of ecosystem function, with the consequent loss of ecosystem services.There remain questions, however, about each of these links As noted above, there arequestions about the link between diversity and ecosystem function Further, the linkbetween ecosystem function and ecosystem services is not well understood In somecases, productivity in an ecological sense is fairly directly linked to value of ecosystemservices in an economic sense For example, the production of biomass ties fairly di-rectly to the provision of forage and the level of carbon sequestration [Tilman, Polaskyand Lehman (2005)] In other cases, the mapping between functions and services may

be quite complex

In sum, there remain a number of unanswered questions about the production andvalue of ecosystem services We currently lack understanding of the link between man-agement actions and changes and ecosystem functioning There remain questions aboutthe link between ecosystem functioning and provision of ecosystem services Finally,there remain questions about the link between the provision of ecosystem services andvalue of these services to humans Provision of valuable ecosystem services may ul-timately prove to be the most important reason to conserve biodiversity At present,however, it is hard to know with any precision either the benefits of the provision ofecosystem services or the opportunity costs involved in ensuring their continued provi-sion.

4 Strategies to conserve biodiversity

A number of human actions cause threats to biodiversity Though over-harvesting is aculprit for some very high profile species (e.g., elephant, rhino, various fish stocks),and climate change poses a threat on the horizon, habitat loss is widely thought to bethe primary cause of biodiversity decline followed by introduction of invasive species[Wilcove et al (1998), Wilson (1992)] We begin this section looking at strategies aimed

at conserving habitat

4.1 Terrestrial habitat protection

Because habitat loss and degradation is viewed as the leading threat to biodiversity,

a major strategy of conservation groups is targeted toward to the goal of preservinghabitat A number of private conservation groups, such as the Nature Conservancy, andgovernment agencies set aside land for biological reserves Additionally, governmentsestablish national parks and other areas where biodiversity is protected Globally, it isestimated that 6.4% of land (excluding Greenland and Antarctica) is in some form ofprotected areas management [UNDP (2000)] Much of the land is set aside because

it is of low value, or because it is of high aesthetic appeal, or for other reasons lated to biodiversity conservation For example, high alpine ecosystems are very wellrepresented in national park systems and wilderness areas, but there are relatively fewecosystems with high agricultural potential conserved

unre-Increasing human population and expansion of economic activity make it unlikelythat habitat loss and fragmentation will cease anytime in the near future, especially indeveloping countries Threats to habitat are generally less in the developed countriesthan in tropical developing countries, where much of biodiversity resides Threats alsovary by type of ecosystem Forests in most developed countries, and some developingcountries are expanding even while tropical rainforests, coastal mangroves and wetlandscontinue to disappear Given limited budgets and facing large threats, conservation plan-ning agencies need to set priorities to ensure that conservation efforts are directed where

they will do the most good Though conservation organizations typically have not lied on the advice of economists, the problem facing these organizations is a classiceconomic problem: how can the greatest conservation return be achieved given limitedresources?

re-One common approach to addressing habitat conservation is to analyze the “reservesite selection problem.” In the standard formulation of the reserve site selection prob-lem, the goal is to choose a collection of reserve sites that among them contain as manyspecies as possible, subject to a constraint on the number of sites that may be included

In other words, the selected sites represent a kind of Noah’s ark Species that inhabit

at least one selected site (in the ark) survive but those that inhabit sites outside of thereserve network (out of the ark) do not This problem of maximizing the coverage ofspecies within a reserve network is what is called a maximal coverage problem in op-erations research[Church and ReVelle (1974), Church, Stoms and Davis (1996)] The

problem can be written formally as follows Define y ias an indicator variable for species

survival: y i = 1 if species i survives and yi = 0 if species i goes extinct, for all i ∈ I,

where I is the set of all species Define x j as an indicator variable for whether site j is selected: x j = 1 if site j is selected in the reserve network and xj = 0 if site j is not

selected, for all j ∈ J , where J is the set of all potential reserve sites The reserve site

selection problem is:

where N i is the set of sites in which species i occurs, and k is the number of sites that

may be included in the reserve network The reserve site selection problem written inthis way is an integer-programming problem that can be solved using linear program-ming based branch and bound algorithms[Csuti et al (1997)]

Much of the early reserve site selection literature did not use optimization methods

to solve for optimal solutions but rather used heuristic methods Two popular heuristicmethods are choosing “hotspots” and the “greedy algorithm” Hotspots are sites thathave the greatest numbers of species The greedy algorithm begins with the site withthe greatest number of species It then picks sites that add the greatest complement

of species to existing reserve sites To see why neither approach guarantees choosing

an optimal reserve network, consider the following simple example fromPolasky andSolow (1999)with four potential sites, labeled A through D, of which only two can be

included in a reserve network Species, which are labeled 1 through 6, that inhabit eachsite are listed in the column for that site (seeTable 1)

A hotspots approach would select sites A and B since they have four species each

while the other sites only have three The greedy algorithm would start by selecting

ei-ther site A or site B and then add eiei-ther site C or site D The optimal reserve network involves choosing sites C and D The hotspots approach fails because it does not con-

sider the complementarity among sites [Pressey et al (1993)] The greedy algorithm,though better than hotspots, fails because it does not allow discarding sites once theyincluded

of species Accumulation curves, which show the number of species included in reservenetworks of various sizes, have a rapid falloff in the accumulation rate after the first fewsites For example,Csuti et al (1997)found that over 90% of all terrestrial vertebratespecies in Oregon were included in a reserve network of five sites, and over 95% wereincluded in 10 sites, but that it took 23 sites to include all species.

The approach outlined in the previous paragraph implicitly assumes that all sites havethe same cost for inclusion in the reserve network It is a simple matter to change from

a constraint on the number of sites,

x j  k, to a budget constraint,
c j x j  B,

where c j is the cost of selecting site j and B is the total conservation budget[Ando et

al (1998), Polasky, Camm and Garber-Yonts (2001)].Ando et al (1998)used this proach to find the minimum cost way of covering endangered species within a reservenetwork in the U.S In an earlier study,Dobson et al (1997)showed that endangeredspecies hotspots occurred largely along the coast of California and in Hawaii Such bio-logical hotspots happen to coincide with real estate hotspots These places include some

ap-of the most expensive real estate in the U.S.Ando et al (1998)show that choosing sitesthat are not necessarily the most biologically rich sites but have a high species per dol-lar ratio is a cost-effective conservation strategy Doing so thereby shifted the focus ofconservation toward the inner-mountain west and away from expensive coastal areas.The cost-effective approach resulted in the same number of endangered species in se-lected sites at one-third to one-half the cost of an approach that included the biologicallyrichest sites regardless of cost

Balmford et al (2003)recently estimated the costs of acquiring reserve sites for diversity protection worldwide They develop a model relates annual management cost[shown byJames, Gaston and Balmford (2001)to be proportional to acquisition cost at

bio-a rbio-atio of bio-about 50 : 1, i.e., bio-a site costing $50 hbio-as bio-annubio-al mbio-anbio-agement costs of bio-about $1]

to variables such as GNP per unit area and purchasing power parity.Balmford et al.(2003)conclude that for acquisition, the highest benefit to cost ratios appear to be al-most exclusively in the developing world

The reserve site selection problem can also be modified to incorporate different jectives besides just maximizing the number of species.Polasky et al (2001)comparedsite selection results when the objective was to maximize phylogenetic diversity versus

ob-maximizing number of species They found that similar patterns emerged, in large partbecause the two objectives are highly correlated Sites with high richness will also tend

to have high phylogenetic diversity and vice versa Adding more species to the mix addsbranch lengths to the phylogenetic tree.Rodrigues and Gaston (2002)showed how to in-corporate phylogenetic diversity measure into an integer-programming problem so thatmaximizing diversity is no more difficult than maximizing species richness They alsofound similar site selection results under richness and diversity objectives In discussedearlier, there are a number of reasons why biodiversity generates both direct and indirectbenefits To date, there have been only limited attempts to tie the objective function ofsite selection problem to the value of conserving various elements of biodiversity

It is also worth noting that most of the literature on conserving biodiversity byprotecting habitat focuses on conserving specific taxonomic groups (e.g., terrestrial ver-tebrates) and does not include all of biodiversity If conserving all of biodiversity is thegoal, the question arises whether conserving a particular taxonomic group is a good sur-rogate for conserving all of biodiversity The conclusion from the biological literature

is that taxonomic groups are not generally reliable proxies for other taxonomic groups

or for all of biodiversity[Prendergast et al (1993), Howard et al (1998), Andelman andFagan (2000)] It may be possible to use environmental surrogates or ecosystem typesrather than species as the units of analysis As yet, however, there is not a solution for agood biodiversity surrogate that commands widespread acceptance

A key simplification of the reserve site selection approach is embedded in the sumptions that species within reserves will survive for sure while those outside of thereserve network will go extinct for sure A more reasonable approach is to model speciesconservation in probabilistic terms One reason for viewing the problem of conservation

as-in probabilistic terms is that there is generally uncertaas-inty about the geographic range

of most species Biologists may only have limited information about whether a givenspecies exists at a given site.Polasky et al (2000)used heuristic methods to find so-lutions to the site selection problem when the goal was to maximize expected speciescovered in a reserve network when there was uncertainty about the geographic range

of species They used the same data on terrestrial vertebrates in Oregon asCsuti et al.(1997)but used probabilities of occurrence The broad geographic pattern of the optimalsolution was similar but there was in general more value to adding similar sites underuncertainty in order to increase the probability that at least one site would contain cer-tain species.Camm et al (2002)showed how to use linear approximations that achieve

a solution arbitrarily close to the optimal solution using linear programming techniquesthat enable rapidly finding the solution of a probabilistic reserve site selection problem.Currently inhabiting a reserve site, however, does not necessarily guarantee the long-term survival of a viable population of the species Survival probabilities can be modeled

as a function of the amount, spatial pattern and quality of habitat Such an approachrequires much more biological information than is required for the reserve site selec-tion problem, which requires only presence/absence data for species at each site Moststudies that incorporate species survival probabilities into a decision analysis aboutwhich lands to protect focus on a single species, largely because of the difficulty of

constructing reasonable spatial population biology models for species Several ies analyzed the optimal spatial pattern of timber harvests to maximize the survivalprobability for a species of concern for a given value of timber harvests, or viceversa [e.g., Hyde (1989), Hof and Raphael (1997), Hof and Bevers (1998), Marshall,Homans and Haight (2000), Rohweder, McKetta and Riggs (2000), Calkin et al (2002)].Montgomery, Brown and Adams (1994)combined timber harvest and demand modelswith a population biology model for the spotted owl to develop an estimated marginalcost curve for increasing survival probabilities They found that the marginal cost in-creased sharply for survival probabilities above 90% Combining population biologymodels with economic analysis allows for greater biological realism as well as allowingfor marginal analysis, showing how slight increases in protected area change survivalprobability and cost.

stud-One of the large challenges facing conservation researchers is how to bridge the gapbetween single species models, which often include detailed biological and economicmodels and data, with the larger breadth and scale of reserve site selection models Con-servation biologists have proposed doing conservation at both a coarse scale and a finescale[Noss (1987)] At the coarse scale, the results of large-scale multi-species analy-sis can direct attention to particular high priority areas for conservation Then a morein-depth fine scale analysis of those particular sites can help develop on the groundconservation plans for those sites Such a two-step process has merit but it does notnecessarily result in optimal or near optimal conservation plans Suppose the fine scaleanalysis of a particular site shows that that conservation is actually more difficult or ex-pensive to accomplish than assumed in the coarse scale analysis If this site is discarded

it may require not just including one different site, but because of the overlap of speciesamong sites, it may change the entire pattern of sites considered to be high priority.Several recent studies have tried to bridge the gap between fine scale and coarse scale

in a single unified analysis.Montgomery et al (1999)analyzed the effects of land use

in Monroe County Pennsylvania on 147 native bird species The study incorporatedinformation on species–habitat associations and territory size to estimate a viabilityfunction for each species based on land use decisions.Lichtenstein and Montgomery(2003)modeled the effect of forestry harvesting decisions on both economic returnsand viability for 166 terrestrial vertebrates in the Coast Range of Oregon.Polasky et

al (2003)have expanded this approach to analyze economic returns and viability forland uses including agriculture, development, forestry as well as conserving land in itsnatural state They find that a strategy that partially integrates conservation within theworking landscape of agricultural and forestry production in addition to having somenatural areas is more efficient than strategies that enforces complete separation of landsinto natural areas, devoted solely to conservation purposes, and production areas, de-voted solely to maximizing economic returns The complete separation of land is anunderlying assumption of the reserve site selection approach

A general question in land management with multiple outputs is whether it is better

to have specialized management, with some land devoted to particular outputs, versusuniform management, with integrated joint production[Boscolo and Vincent (2003)]

Nonconvexities in production may give rise to specialization One simple example of anonconvexity in the joint production of timber and biodiversity occurs with species thatrequire large tracts of undisturbed forests The first units of timber production may result

in a fall in the species population to near zero Further increases in timber productionwould not therefore affect the species population significantly In this case, it is better

to have some undisturbed tracts with intensive timber production elsewhere rather thannon-intensive timber production everywhere Forestry models with joint production ofbiodiversity and timber typically find that it is optimal to have at least some degree ofspecialization[Bowes and Krutilla (1989), Swallow, Parks and Wear (1990), Swallowand Wear (1993), Vincent and Binkley (1993), Swallow, Talukdar and Wear (1997),Boscolo and Vincent (2003)] Similar results on specialization have been found for al-locating conservation funds across different areas[Wu and Boggess (1999), Wu, Adamsand Boggess (2000)]

Another challenge to conservation researchers is to include dynamics into the sis Development activity, climate change, biological invasions, changes in relativeprices, and chance events such as forest fires and drought may all play a role in chang-ing the landscape that require changes in conservation priorities and an ability to adapt.Work is beginning to appear that combines predictions of regional climate change and itsimplications for conservation[Parmesan and Yohe (2003), Root et al (2003)].Costelloand Polasky (2003)analyze a dynamic reserve site selection problem in which each sitethat is not protected has a probability of being developed during that period A conserva-tion agency facing a budget constraint each period must choose sites knowing that it willget to choose again in the future, but also knowing that not all remaining high priorityconservation sites will remain undeveloped.Meir, Andelman and Possingham (2003)analyze a similar problem except that in their paper the probability is whether a site willbecome available during the period When a site becomes available it must either bepurchased by the conservation agency that period or face a development risk For thesetypes of problems, the optimal sites to choose are those that combine high biologicalvalue added per unit cost and face a high development threat This conclusion providessome support for the strategy of giving high priority to conserving biodiversity hotspots,defined as areas of high biodiversity or high endemism that face large threats of habitatloss [e.g.,Mittermeier et al (1998), Myers (1988), Myers et al (2000)].Kareiva andMarvier (2003)criticize this approach for ignoring other important issues in conser-vation, such as cost and other conserving objectives beyond species (e.g., ecosystemfunctions and services) As noted above, hotspots will fail to be optimal if it does notfactor in complementarity among sites Many of these criticisms can be addressed cor-rectly specifying the objective function and the set of constraints For any dynamicconservation problem, biological value added, cost and threat will be essential elementsthat will drive the analysis

analy-There is a fast-growing literature analyzing habitat conservation issues Some work

in this vein has begun to incorporate both biological and economic objectives in an grated fashion A key challenge, though, is to be able to incorporate realistic economicand biological details while maintaining the broad-scale focus necessary for setting

inte-biodiversity conservation priorities At present, most models are too stylized to be ofuse in helping decision-makers choose conservation strategies on the ground, thoughthe gap between models and reality is closing Another challenge in habitat conserva-tion models is to incorporate dynamics and stochastic events in the analysis.

4.2 Marine biodiversity and reserves

While similar in many respects to the terrestrial environment, marine systems present

a host of challenges for biodiversity conservation and reserve design Ecologists warnthat marine biodiversity, and the coastal ecosystems on which it depends, is in peril

A recent paper cites a precipitous decline in many marine consumer species from tastically large” historical levels[Jackson et al (2001)] Analyzing data beginning inthe 1950s,Myers and Worm (2003)find that the global ocean has lost more than 90%

“fan-of large predatory fish, and that industrialized fisheries have typically reduced munity biomass by 80% since large scale exploitation began approximately 50 yearsago Although this collapse has been attributed to several anthropogenic sources such

com-as pollution, degradation of habitat quality, and changes in climatic regimes, it is man overexploitation of marine resources that tops the list of threats[Jackson et al.(2001)] The traditional approach to managing marine resources – with single speciesmanagement and little account of uncertainty, learning, or dynamic ecological effects[popularized for example by the early models ofGordon (1954), Smith (1969), amongothers] – has largely failed And an emerging body of literature makes it clear that newapproaches to the management of marine resources will be required if biological diver-sity is to be protected

hu-In response to this call, attention has recently shifted from single-species policy sign to the implementation of marine reserves.Botsford, Hastings and Gaines (2001)argue that perhaps 35% of marine ecosystems must be set aside in order to protectbiodiversity Currently only about 0.5% is protected[IUCN (1997), Kelleher, Bleakleyand Wells (1995)] In addition, some claim that by setting aside large tracts of produc-tive marine ecosystems, we may earn a “double payoff” through enhanced biodiversitywithin the reserve and increased fishery yield from spillovers outside the reserve, thoughthis claim is subject to some dispute[Sanchirico and Wilen (2001)]

de-Several key features distinguish the design of marine reserves from the design of restrial reserves Important differences include the biology of target organisms, threats

ter-to native biodiversity, and political economy considerations Many marine organismsare characterized by planktonic larval stages, where larvae may drift with ocean cur-rents for several weeks, often traveling hundreds of miles before settling And in theiradult stages, many marine species are highly mobile This high level of mobility dur-ing several life-history stages has at least two important implications for marine reservedesign First, even if marine reserves are sited in the most productive areas, small (con-tiguous) reserves may not protect biodiversity as effectively as a reserve around sessilespecies [see, e.g.,Botsford, Hastings and Gaines (2001)] And second, there may ex-ist a “spillover” effect where larvae produced within a reserve spill into adjacent areas