Optimal Fishing Policies That Maximize Sustainable Ecosystem Services

Home , Sustainable yield

K. Tsukamoto, T. Kawamura, T. Takeuchi, T. D. Beard, Jr. and M. J. Kaiser, eds. Fisheries for Global Welfare and Environment, 5th World Fisheries Congress 2008, pp. 359–369. © by TERRAPUB 2008.

Hiroyuki Matsuda,1* Mitsutaku Makino2 and Koji Kotani3

1Faculty of Environment and Information Sciences Yokohama National University 79-7, Tokiwadai, Yokohama 240-8501, Japan

2National Research Institute of Fisheries Sciences Fisheries Research Agency 2-12-4, Fukuura, Yokohama 236-8648, Japan

3Graduate School of International Relations International University of Japan 777 Kokusai-cho, Niigata 949-7277, Japan

*E-mail: [email protected]

The classic theory of fisheries management seeks a maximum sustainable yield from a target species. There are several variations that include uncertainty, fluctuation, and species interactions. The knowledge that sustainable fisheries do not always guarantee conservation of a diversity of species is well known. Ecosystems provide several categories of ecosystem services to human well- being: supporting, provisioning, regulating, and cultural services. Fishery yields belong to provisioning services. The existence of living marine resources may maintain these services, and certainly a much larger contribution from regulating services than that from fishery yields. Therefore, we define an optimal fishing strategy that maximizes the total ecosystem service instead of a sustainable fishery yield. We call this the fishing policy for “maximum sustainable ecosystem service” (MSES). The regulating service likely depends on the standing biomass, while the provisioning service from fisheries depends on the catch amount. We obtain fishing policies for MSES in a single species model with and without uncertainties and in multiple species models. In any case, fishing efforts are usually much smaller than those for a maximum sustainable yield (MSY). We also discuss the role of fisheries in sustaining ecosystem services, and the nature of ecosystem comanagement.

KEYWORDS maximum sustainable ecosystem service; MSES; uncertainty; ecosystem comanagement; maximum sustainable yield 360 H. MATSUDA et al.

able yield under which all species coexist. 1. Introduction However, Matsuda and Abrams (2006) did not incorporate ecosystem service into the The theory of maximum sustainable yield optimal fisheries policy. (MSY) takes into account the long-term yield In this paper, we consider an optimal from living marine resources. This theory policy that maximizes the total ecosystem implicitly assumes that a fishing ban does service under some mathematically simple not produce any benefit from marine eco- assumptions. We call this the maximum sus- systems, although this theory explicitly as- tainable ecosystem service, or MSES. We sumes a negative relationship between yield assume that an ecosystem service other than and standing stock abundance. However, fishery yields depends on the standing ecosystems give us a variety of benefits biomass, while the fishery yield depends on (World Resource Institute 2005). Harvests the catch amount. Regulation of fishing ef- from agriculture, forestry and fisheries are forts usually enhances the standing biomass, just a small part of the ecosystem service while it usually decreases the fishery yield. (Costanza et al. 1997; Satake and Rudel We compare the MSY and MSES policies in 2007). Ecosystem services include support- (1) a single deterministic stock dynamic ing services such as soil formation, photo- model, (2) a stochastic model for a single synthesis, and nutrient cycling, provisioning species with measurement errors and proc- services such as food, water, timber, and ess uncertainty and (3) food web models fiber; regulating services that affect climate, consisting of six species. We also discuss the floods, disease, waste, and water quality; and role of comanagement in the ecosystem ap- cultural services that provide recreational, proach. aesthetic, and spiritual benefits (World Re- source Institute 2005). Fishery yields belong 2. Optimal Fishing Policy That to provisioning services. The existence of liv- Maximizes Ecosystem Service ing marine resources may maintain these services and certainly make a larger contri- To take account these ecosystem services, we bution to the total ecosystem service than assume that the total ecosystem service from fishery yields (Costanza et al. 1997). a target fish resource, denoted by V(N, C) is There are some criticisms of the theory given by of MSY (Matsuda and Abrams 2008). The MSY fishing policy does not reflect uncer- V(N, C) = Y(C) – cE + S(N) (1) tainty in stock estimates (measurement errors) and in the relationship between the where Y(C) is the yield from fisheries with spawning stock and recruitment (process catch amount C, cE is the cost of fisheries uncertainties). The MSY policy also ignores with fishing effort E, and S(N) is the ecosys- the complexity in ecosystem processes be- tem service other than fishery yields with cause the theory uses single stock dynamic stock abundance N. Hereafter, we simply call models (Matsuda and Katsukawa 2002). S(N) the utility of standing biomass. Matsuda and Abrams (2006) analyzed the We assume the following fish stock dy- maximum sustainable yield from entire food namics : webs with an independent fishing effort for each species. They concluded that MSY dN/dt = f (N)N – C, policy does not guarantee the coexistence of f (N) = r – aN, species and proposed the concept of “con- C = qEN, strained MSY” that maximizes the sustain- Y(C) = pqEN (2) Optimal fishing policies 361 where t is an arbitrary time unit, f(N) is the per capita reproduction rate, r is the intrinsic rate of population increase, a is the magnitude of density effect, q is catchability, and p is the price of fish. Although we have assumed linear functions for cost (cE) and yield (pqEN), the nonlinear relationship for either cost or catch may produce a more complex result. The carrying capacity, denoted by K, is given by r/a. The utility of standing biomass, S(N), is Fig. 1. The relationship between stock abun- likely to be a convex or sigmoid curve be- dance and its utility given by Eq. (3). Bold and cause the utility usually saturates when the thin curves represent the cases where B is 50 stock is sufficiently abundant. We assume the and 10% of K, respectively. following specific function:

S(N) = S∞N2/(B2 + N2) (3) where S∞ is the limit of S(N) when N is the positive infinity, B is the stock abundance where S(N) is the half of S∞. We assumed a sigmoid function of N as shown in Fig. 1. In the first step we obtain the optimal fishing effort that maximizes the total ecosystem service V(N, C) given by Eq. (1) at equilibrium stock abundance of Eq. (2). The equilibrium stock abundance, denoted by N*, is given by Fig. 2. The relationship between the fishing effort and the total ecosystem service for three N* = (r – qE)/a (4) cases of parameters. Bold, thin and dotted curves represent the cases where (S∞, B) = (100, 10), The total ecosystem service at the equilib- (100, 50), (0, —), respectively. Other parameters are chosen as = 0.01, r = 1, q = 1, p = 1, c = 0.1. * a rium, denoted by V , is The optimal effort for each case is indicated by closed circles. V* = pqEN* – cE + S(N*) (5)

If the utility of the standing biomass is negligible (S(N) = 0) as is usually assumed If S(N) is an increasing function of N, in classical fisheries theory, V* is a uni-mo- the optimal effort is always smaller than Eopt dal function of E and the optimal fishing ef- without the utility of standing biomass. How- ever, the distance between E with positive fort, denoted by Eopt, is well known by opt S(N) and Eopt without S(N) depends on the 2 magnitude and curvature of S(N). We nu- Eopt = (pqr – ac)/2pq (6) merically obtained the optimal effort for vari- ∞ If cost c is negligible, Eopt becomes the ef- ous parameters of S and B as shown in fort at the maximum sustainable yield Fig. 2. Despite the fact that the maximum * * * ∞ (MSY). The quantity V when E = Eopt is Vopt(N , C ), when (S , B) = (100, 50) is larger * * ∞ known as the maximum economic yield. than Vopt(N , C ) when (S , B) = (100, 10), 362 H. MATSUDA et al. the latter optimal effort is smaller than the ment error ζ(t). In this case, the fishing ef- former. fort Et likely depends on time because the The derivative of V* with E is stock fluctuates with process errors. In the decision making, fishers just use the estimate * * * ˜ ˜ dV /dE = pqN – c – [pqE + s′(N )]q/a (7) N t, therefore Et is a function F of N t. We here assume one of the simplest ways because dN*/dE = –q/a. that is similar to Reed (1979): The optimal effort is smaller when the derivative of utility of the standing resource f (Nt, ξt) = exp(r + ξt – aNt) (9) with respect to the stock is of a larger magnitude (|dS/dN| is larger). If S∞ > (Ba2 + We also incorporate measurement uncer- r2)2(rpq – ac)/2Bqra3, dV*/dE at E = 0 is nega- tainty: tive and the fishing ban is optimal. For parameter values given in Fig. 2, the fishing G(Nt, ζt) = Nt exp(ζt) (10) ban is optimal when S∞ > 47595 if B = 10 or when S∞ > 9995 if B = 50. The optimal harvesting policy is obtained by the dynamic programming theory 3. Optimal Fishing Policy with (Mangel and Clark 1988) although the solu- Process Uncertainty and tion is still analytically unknown if a meas- Measurement Errors urement error exists (Kotani et al. 2008). Here we simply assume that the allowable Equation (1), that includes the value of effort rule in actual fisheries management in ecosystem service, is simple because any the USA, Japan, and other countries as types of uncertainties, intrinsic instability, shown in Fig. 3: complexity and conflict between stakehold- ers are ignored. We incorporate these fac- ⎧0 if NN˜ < tors into the model to build the comprehen- ⎪ t ban sive theory of fisheries management. In the ⎪ ⎪ ˜ second step we consider a time-discrete ⎪FNNNtarget()− ban() limit− N ban FN()˜ = ⎨ model with the uncertainties in measurement ˜ ⎪ if NNNban≤ t < limit and dynamic processes: ⎪ ⎪ ˜ ⎪FNNtargetif t ≥ limit Nt+1 = f (Nt – Ct, ξt)(Nt – Ct) ⎩ ()11 Y(Ct) = pCt = pqEtNt ˜ N t = G(Nt, ζt) where Nban and Nlimit are the stock abundances V(Nt, Ct) = Y(Ct) – cEt + S(Nt – Ct) of the upper threshold for a fishing ban and ˜ Et = F( N t)/q (8) the lowest threshold that maintains a fishing effort Ftarget, Ftarget is the target fishing effort where Nt is the stock abundance in year t just under which the maximum sustainable yield before the fishing season, ξt and ζt are nor- is theoretically achieved. The conventional mally-distributed random variables with fisheries management under the United Na- standard deviations σr and σm, f(Nt – Ct, ξt) tions Convention on the Law of the Sea is the per capita reproduction rate with (UNCLOS) usually determines the total al- ˜ process uncertainty ξ(t), N t is the estimate lowable catch (TAC) instead of the fishing of stock abundance Nt in year t that is a func- effort. Therefore, the actual catch should N˜ tion G of the true abundance Nt and measure- have been the minimum of qEtNt and qEt t. Optimal fishing policies 363

Fig. 3. Rules of fishing effort that depend on the estimate of stock abundance. Bold, thin and dotted curves represent the ABC rule given by Eq. (11), constant escapement given by Eq. (12)

and constant catch given by Eq. (13), where q = 1, Ncrit = 100, Cc = 200, Nlimit = 600, Ftarget = 0.5.

Here we simply assumed that Ct is given by larger average yield and a larger minimum qEtNt or that the control of fishing effort de- stock with a smaller variance. This solution pends on the estimate of stock abundance implicitly incorporates the utility of stand- (see also Katsukawa 2004). The fishing mor- ing biomass into the performance of fishing tality coefficient F is approximately given policy. Katsukawa (2004) concluded that the by qE. fishing effort that achieves the maximum If we ignore the utility of standing sustainable yield is a bad performance. He biomass (S(N) = 0) and measurement error recommends a simpler rule given by Eq. (11)

(σm = 0), the constant escapement policy is that is similar to the New Management Pro- well known as the optimal policy that maxi- cedure of the International Whaling Com- mizes the long-term yield: mission. Kotani et al. (2008) considered the optimal control of exotic species and numeri- ⎧0 if NNt < crit ⎪ cally obtained an optimal harvesting policy EN()= cs ⎨ when the utility of standing population is ⎪ ⎩()NN− crit qNif Nt ≥ N crit negative. Even for the optimal control of ()12 exotic species, the constant escapement is optimal. where Ncrit is a positive constant. The catch Figure 4 illustrates a resultant stock qEt,optNt is (Nt – Ncrit) and the stock abundance abundance, catch amount from MSY, and after the fishing season is Ncrit. The optimal MSES policies. We obtained the optimal escapement level depends on the magnitude policies by grid search for Ftarget, Nban, Nlimit of process uncertainty (σr). The fishing ef- as 0.1, 100, 100, respectively, to maximize fort for the constant catch policy is the average yield or ecosystem services over 100 years from 100 simulations. The fishing

Ecc(N) = Cc/qN (13) effort under the optimal policy for MSY was higher than that for MSES, while the stock where Cc is the constant catch amount. abundance under the optimal policy for MSY

If the measurement error exists (σm > 0), was lower than that for MSES. We chose ∞ Katsukawa (2004) numerically obtained a parameters (S , B, r, a, σr, σe, p, c) = (100, suitable management policy that keeps a 1000, 0.5, 0.001, 30%, 50%, 1, 0). We obtained 364 H. MATSUDA et al.

Fig. 4. An example of stock and catch fluctuation under fishing effort given in Eq. (11). Policies for MSY and MSES are shown in panels (a) and (b), respectively. Bold and dotted curves represent the stock abundance and fishing effort. Circles represent the catch.

a fishing policy characterized by (Ftarget, Nban, Matsuda and Abrams (2006) did not ex-

Nlimit) in Eq. (11). The policy for MSY was plicitly evaluate the ecosystem service of

(Ftarget, Nban, Nlimit) = (0.6, 100, 700), whose standing biomass but added a constraint of average yield and average total ecosystem species persistence. Here we incorporate the services were 77.6 and 165.7, respectively. utility of ecosystem service from standing

The policy for MSES was (Ftarget, Nban, Nlimit) biomass into their model that includes s spe- = (0.2, 100, 1000), whose average yield and cies: average total ecosystem services were 43.3 and 216.4, respectively. dNi/dt = (ri + ∑ajiNj – qiEi)Ni, for i = 1, 2,..., s (14) 4. Optimal Policy from Food Webs

Ecosystems are characterized by uncer- where Ni(t) is the stock abundance of species i at time t, r is the intrinsic growth rate tainty, dynamic properties and complexity. i of population increase of species i, a is the We have analyzed the effects of uncertainty ji magnitude of intra- and inter-specific com- and dynamic properties previously. Matsuda petition from species j to i, and E is the fish- and Abrams (2006) used simple food-web i ing effort on species i. models to investigate the nature of yield- or We define the total ecosystem service profit-maximizing exploitation of commu- from the food web: nities including a variety of six-species systems with as many as five trophic levels. Y(E) = E (p q N – c ) These models show that, for most webs, rela- ∑ i i i i i (15) tively few species are harvested at equilib- V(E) = Y(E) + S (N ) rium and that a significant fraction of the ∑ i i species is lost from the web. They also con- where p, q, c and S differ between species. sidered a constraint that all species must be Matsuda and Abrams (2006) ignored the util- retained in the system usually increases the ity of standing biomass S (N ). There may be number of species and trophic levels har- i i some multiplicative effects of interspecific vested at the yield-maximizing policy. The interactions on the ecosystem service. It is reduction in total yield caused by such a con- difficult to quantitatively evaluate the total straint is modest for most food webs. ecosystem service from the web. Here we Optimal fishing policies 365 assumed the simple sum of the contribution ity of 50% and its absolute value is between of each species on the total ecosystem serv- 0 and 1 with a probability of 50% for any ice. pair of predator species i and prey species j

(j < i). We also assumed that qi is 1, because ∞ 2 2 2 Si(Ni) = Si Ni /(Bi + Ni ) (16) q and p similarly affect the optimal solution. Other parameter values are chosen by inde- We evaluate the total ecosystem service pendent draws from a uniform distribution at the equilibrium. The equilibrium, denoted between 0 and 1 for ri, p1 and p2, and be- * * * * T by N = (N1 , N2 , ..., Ns ) , is given by tween 0 and 3 for p3 to p6 because predators usually fetch a higher price than prey spe- NArqE∗−= −−1() cies. We ignored the cost of fishing effort, ∞ or so ci = 0. Finally, we assumed that Si and Bi ** are 10 and 0.2Ni for each species, where ∗ L −1 ⎛ N ⎞ ⎛ aaa11 12 13 a 1s ⎞ ⎛ rqE111− ⎞ ** 1 Ni means the equilibrium stock abundance ⎜ ∗⎟ ⎜ L ⎟ ⎜ ⎟ N2 aaa21 22 23 a 2s r2 − qqE22 without fisheries (E = 0). ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ∗ L ⎜ N3 ⎟ = ⎜aaa31 32 33 a 3s ⎟ ⎜ rqE333− ⎟ Figure 5 shows four typical examples ⎜ M ⎟ ⎜ MMMOM⎟ ⎜ M ⎟ of food webs. When the total ecosystem serv- ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ∗⎟ L ice decreases with an increasing fishing ef- ⎝ Ns ⎠ ⎝ aaass123 s a ss⎠ ⎝ rqEsss− ⎠ fort of any species, the MSES policy does () 17 not fish any species at all, as shown in Fig. 5(a). Among 1000 randomly connected food where r = (r , r , r , ..., r )T; qe = (q E , q E , 1 2 3 s 1 1 2 2 webs, about 85% of the webs have the same q E , ..., q E )T and A means the community 3 3 s s food webs with the same set of exploited matrix, the (i, j)-th elements of which is a . ij species. About 15% of the webs have the We obtained two types of solutions of same food webs with the same fishing ef- fishing efforts E that maximize Y(E) and forts between MSY and MSES (Fig. 5(b)). V(E) in Eq. (15) for the same food web with The total yield at MSES (denoted by cost and prices, A, c, q, p, r. We call these Y ) ranged 0–100% of the total yield at solutions “MSY policy” (E ) and “maxi- MSES MSY MSY (denoted by Y ) whose average was mum sustainable ecosystem service (MSES) MSY 40%. The total ecosystem service at MSY policy” (E ), respectively. We compare MSES (V ) ranged between 21 and 100% of the between the total ecosystem service V(E) at MSY total ecosystem service at MSES (V ) MSY andV(E) at MSES, denoted by V and MSES MSY whose geometric average was 69%. There was V , respectively. The total yield using the MSES a positive relationship between Y /V MSY policy is always larger than that for MSY MSES and Y /Y (Fig. 6), while a fishing ban the MSES solution, while the total ecosys- MSES MSY at MSES appeared in 1% of the 1000 food tem service for the MSES solution is always webs when the fishery yields were the larger than that for the MSY solution. smaller part of the total services (Y /V We obtained MSY and MSES policies MSY MSES is small) and the MSES policy was identical for 1000 randomly structured food webs of to the MSY policy when the fishery yield a six species food web with positive equi- was the larger part of the total services. librium. We assumed that the diagonal ele- All species rarely coexisted under the ments aii are –1 for bottom-level species MSY policy for 1000 randomly structured (i = 1 and 2) and 0 for consumer species (i = food webs, while these six species coexist 3 through 6). We ignored interspecific com- for about 91% of the webs using the MSES petition between species 1 and 2 (a = a = 12 21 policy. The number of exploited species at 0). We also assumed a is 0 with a probabil- ij MSES was usually much larger than the 366 H. MATSUDA et al.

Fig. 5. Resultant food web and optimal fishing policies that maximize the total sustainable yield (panels (a)–(d)) and the total ecosystem service (panels (e)–(h)). The MSES policy for four examples of the six-species systems; circles and arrows represent species and fishery, respectively. Lines between circles represent trophic links from smaller numbered species to larger numbered species. Dotted circles mean that these species became extinct. Panels (e)–(h) are the MSES policies for four systems that are identical to panels (a)–(d), respectively.

Fig. 6. The relationship between the relative ecosystem service and the relative total yield under the MSES policy from 1000 random food webs. Optimal fishing policies 367

Table 1. Resultant food webs and fishing efforts from 1000 randomly constructed six species systems. Each column shows the frequency distribution of the numbers of extant species using MSY and MSES policies, the number of exploited species using MSY and MSES policies.

No. of extant species No. of exploited species No of extant species at No of exploited species at MSY at MSY MSES at MSES

1 31 160 0 0 2 139 828 178 201 3 48 10 431 512 4 56 0 220 256 5 18 0 24 28 6 6 0 144 0

number at MSY (Table 1). The ratio YMSY/ of a significant fraction of species in the web.

VMSES implies the ratio between the maximum It will inevitably lead to the degradation of sustainable yield and the maximum sustain- S(N), and easily set off the benefit from able ecosystem service by respectively opti- Y(E*), and ultimately reduce the total eco- mal policies. This ratio widely ranged be- system services, V(E). To avoid these situa- tween 0 and 97%. If the total ecosystem serv- tions, government has to monitor the rest of ice without a fishery yield is small, the MSES the ecosystem, and regulate the yield-maxi- policy did not guarantee the coexistence of mizing fisheries in a top-down way. The re- species. ality is, again, these costs would be beyond the budget of many countries, especially de- 5. From Fisheries Comanagement to veloping countries. To sum up the above dis- Ecosystem Comanagement cussions, yield-maximizing, economically- efficient fisheries are rational for enjoying In order to increase the fishery yield fishery rent, but not always so in sustaining Y(E), monitoring activities of status of stock total ecosystem services for society. are critically important. Even more so for From the viewpoint of sustaining eco- sustaining the utility of standing biomass system services, one reasonable alternative S(N). Government should play an important is to conduct responsible fisheries targets for role in these monitoring activities. However, a wide range of species with a variety of gear. in reality, it is almost impossible for the gov- For example, in the Shiretoko World Natu- ernment to monitor all the detailed ecosys- ral Heritage area, local fishers have con- tems along the coast and within exclusive ducted various kinds of operations and economic zones (EEZ). Therefore, the caught most of the keystone species of the knowledge of fishers and data from fishery local ecosystem. They have accumulated the activities should be fully utilized. catch data of these species for over 50 years. As Table 1 shows, yield-maximizing In addition, local fishers have autonomously fisheries are likely to take only a small monitored the resource status every year, and number of species from one trophic level. adaptively modified the local rules for sus- So, under the yield-maximizing policy, fish- tainable resource use. These data and knowl- ing gear and vessel type will converge on edge is now one of the most important foun- the most efficient, and we can gain informa- dations for monitoring the changes in the tion about very limited aspects of the eco- ecosystem structure and functions in the system. Our analyses also indicate that yield- heritage area. As this case shows, responsi- maximizing fisheries would invoke the loss ble fisheries can significantly contribute to 368 H. MATSUDA et al. the sustainability of ecosystem services. Like system processes, including food web inter- the Shiretoko case, a fisheries management action, are complex and not well known. The approach in which government and local assumptions and parameter values used in fishers share the responsibilities and authori- this paper should be improved in the future. ties for the use of sustainable resource is Our analyses have the following problems. called fisheries comanagement, the strong- First, we do not know the appropriate ∞ est argument against the conventional top- magnitude of ecosystem services (Si ). We down approach (Makino and Matsuda 2005). assumed that the ecosystem services from a Fisheries comanagement will not always standing biomass is a simple sum of the serv- lead to the ecosystem management. As this ice from each species with the same magni- ∞ study showed, the total ecosystem service tude of Si . Even for such simple models, we from MSES policy is always larger than that obtained a variety of MSES policies from a from MSY policy, and EMSY has a tendency fishing ban of all species to the extinction of to be greater than EMSES. In order to increase some species. The regulating services are at the total ecosystem services, interests from least 10-fold larger than fishery yield other sectors than fisheries, such as an envi- (Costanza et al. 1997). Our analyses suggest ronmental ministry or non-government or- that a fishing ban is optimal, even when the ganizations (NGOs), should be included in fishery yield using the MSY policy is much the decision-making arena (Makino 2005). more than 10% of the total ecosystem serv- Then, fisheries comanagement will evolve ices using the MSES policy. If the regulat- into an ecosystem comanagement. Under this ing services do not significantly decrease ecosystem a comanagement framework, co- through little fishing effort, a fishing ban is existence of small-scale, artisanal fisheries less likely, but a significant reduction of the and large-scale, efficient fisheries is the ra- standing biomass is strongly discouraged. tional solution for sustaining ecosystem serv- Second, we do not include implementa- ices. tion errors. This is a big problem in consen- sus building. Depending on the countries or 6. Discussion areas, food supply and job creation by greater fishing efforts are considerably important In the three types of models we have factors from a social security point of view. analyzed in this paper, the concept of maxi- Therefore, relative weights between ecosys- mum sustainable ecosystem services results tem services and fishery yields might be the in a more conservative policy than MSY. A societal choice. fishing ban is optimal if the ecosystem service from standing biomass is sufficiently Acknowledgement larger than the fishery yields, despite the fact that MSES do not always prevent any spe- We thank Dr. A. Satake and members of the Global cies from becoming extinct. These results are COE Program “Eco-Risk Asia” at Yokohama National intuitively understandable. University and National Institute for Environmental It is difficult to quantitatively evaluate Studies for valuable comments and encouragement. the magnitude of ecosystem services. Eco- This work is in part supported by a JSPS grant to H.M.

References

Costanza R, d’Arge R, de Groot R, Farber S, Grasso M, Hannon B, Limburg K, Naeem S, O’Neill RV, Paruelo J, Raskin RG, Sutton P, van den Belt M. The value of the world’s ecosystem services and natural capital. Nature 1997; 387: 253-260. Optimal fishing policies 369

Katsukawa T. A numerical investigation of the optimal control rule for decision-making in fisheries management. Fish. Sci. 2004; 70: 123–131. Kotani K, Kakinaka M, Matsuda H. Optimal escapement levels on renewable resource management under process uncertainty: Some implications of convex unit harvest cost. Env. Econ. Pol. St. 2008; 9: 107– 118. Makino M. The Ecosystem Approach of the CBD and fisheries management in Japan. Glob. Env. Res. 2005; 7: 95–104. Makino M, Matsuda H. Co-Management in Japanese Coastal Fishery: It’s Institutional Features and Transac- tion Cost. Mar. Pol. 2005; 29: 441–450. Mangel M, Clark CW. Dynamic Modeling in Behavioral Ecology. Princeton Univ. Press, Princeton. 1988. Matsuda H, Katsukawa T. Fisheries management based on ecosystem dynamics and feedback control. Fish. Oceanogr. 2002; 11: 366–370. Matsuda H, Abrams PA. Maximal yields from multi-species fisheries systems: rules for systems with multiple trophic levels. Ecol. Appl. 2006; 16: 225–237. Matsuda H, Abrams PA. Can we say goodbye to the maximum sustainable yield theory? Reflections on trophic level fishing in reconciling fisheries with conservation. in Nielsen JL, Dodson JJ, Friedland K, Hamon TR, Musick J, Verspoor E (eds). Reconciling Fisheries with Conservation. Proceedings of the Fourth World Fisheries Congress. American Fisheries Society, Symposium 49, Bethesda, Maryland. 2008; 737–744. Reed WJ. Optimal escapement levels in stochastic and deterministic harvesting models. J. Env. Econ. Manag. 1979; 6: 350–363. Satake A, Rudel TK. Modeling the forest transition: forest scarcity and ecosystem service hypotheses. Ecol. Appl. 2007; 17: 2024–2036. World Resource Institute (ed). Millennium Ecosystem Assessment: Ecosystems and Human Well-Being: Syn- thesis. Island Pr., St. Louis. 2005; 137 pp.