1 Importance of Type-II Error and Falsifiability Hiroyuki MATSUDA Univ. of Tokyo, IWC/SC Japan Delegate WWF Japan Committee Member This Powerpoint file will be uploaded on tokyo.ac.jp/~matsuda/2002/ ppt

2 Precautionary principle Rio Declaration 1992, Principle 15 “In order to protect the environment, the precautionary approach shall be widely applied by States according to their capabilities. Where there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a reason for postponing cost- effective measures to prevent environmental degradation.

3 Convention on Biological Diversity JUNE 1992 “Noting also that where there is a threat of significant reduction or loss of biological diversity, lack of full scientific certainty should not be used as a reason for postponing measures to avoid or minimize such a threat,

4 UN Framework Convention on Climate Change “Where there are threats of serious or ir- reversible damage, lack of full scientific certainty should not be used as a reason for postponing such measures, taking into account that policies and measures to deal with climate change should be cost-effective so as to ensure global benefits at the lowest possible cost.

5 No academic rule for what we should say has been established. Scientists … Before the Earth Summit in 1992, –should give no comments to public without full scientific evidence; –keep their result irrespective of public opinion After must give some make their opinion a public consensus or win votes Galileo’s Inquisition

11 IUCN Redlist Criteria (2001)

12 Risk analysis is based on a threshold of type-II error. Type-II: The probability that a species goes extinct when it is not listed as endangered. Type-I: The probability that a species persists when it is listed as endangered (very small). Or, If the extinction risk of a species is >10% within next 100 years, it is listed as endangered.

14 G. Mace et al : Species 19:16. (The validity of criterion A:) “ it can result in the listing of some species with very large, apparently secure populations ”. (Type-I error) “ However, linking [the rates of de- cline] to population size would exclude the listing of many populations with limited census data. ” (Type-II error)

15 Mrosovsky N (1997) Nature 389:436

16 Allow criterion E to over-rule other criteria !? If we do not evaluate extinction risk, we agree with listing a species by criteria other than Criterion E. We disagree with listing it by criteria A-D if estimated extinction risk is apparently low. No consensus was made in IUCN Marine Workshop. About 2/3 of IUCN Criteria Workshop participants disagreed with this option.

Risk is usually evaluated under pessimistic assumptions. IUCN/SSC (p.25) “Assessors should resist an evidentiary attitude and adopt a precautionary but realistic attitude to uncertainty when applying the criteria, for example, by using plausible lower bounds, rather than best estimates, in determining population size...” Therefore, extinction risk based on pessimistic estimates is biased (-fit to avoid type II errors) We could take risk based on best estimates, and measure type I errors (the weight of evidence).

Japanese plant Red Data Book Questionnaires: The number of plants and decline rate in each of 4437 map grids in each of ca.2100 threatened? plant species. Calculate total population size, rate of population decline, extinction risk of each species. >1500 species are listed in RDB.

Frequency distribution of grids The case of primura sieboldii extinction13 N p =f 1 N 1 + f 2 N 2 + f 3 N 3 + f 4 N 5 =31977 >1000<0.01<0.1<0.5<1>1?total > > > > ?12223 total Decline rate within past 10 years Population size 23 Pessimistic assumption = ignoring unknown grids Unbiased assumption = proportional divide

We define the weight of evidence in plant RDB Extinction risk: based on pessimistic assumptions (ignoring unknown grids) Weight of evidence: based on unbiased assumptions (proportional divide) For 8 CR taxa, 32 EN taxa and 14 VU taxa among 1325 taxa, the weight of evidence within the years in question did not satisfy the extinction risk criteria.

The weight of evidence decreases with increasing number of size unknown grids EN? VU? NT?

How to handle the weight of evidence… We do not need down-listing even in case of disagreement between scenarios with pessimistic and unbiased estimates (PP) We should show the weight of evidence for future review process (accountability) Like weather focast (risk of shower)

6 Fallacy of applying PP to Maximum Sustainable Yield

8 Threat of biodiversity is serious if the population is below MVP Minimum viable population (MVP) is defined as threat of demographic stochasticity (e.g., all mothers make sons = 50) and genetic degradation (=500). The “50/500” law does not guarantee a zero-risk. If population size > 10,000, the mean time to extinction is usually far too long (I ignore > 1 million yrs).

9 Should any few risk be avoided? (IWC 2001 report, p.93) Exploitation of whales with environmental variability was still “equivalent to an unsustainable ‘mining’” (still positive risk) Under the RMP, “the time scales were far too long (10 45 years)” (>>the age of cosmos) “ the long time-scale was necessary to examine the mechanisms of the interaction between environmental variability and exploitation. ” ????

RMP cares just 10% errors Production Catch quota Fishing rate MSY limit stock level MSY at 60%, 0 catch at 54%

Fisheries Management Rule I US and Japan No lower stock limit Fishing is possible until stock collapse Uncertainty exists not only in whaling, but all fisheries.

10 Fallacy of applying PP to MSY If we adopt biased (precautionary) estimates, expected yield is again negatively biased. MSY should be based on unbiased, most likely estimates (Error in quota is reversible = adaptive management) MVP should be based on biased estimates, or PP. (Lost of biodiversity is irreversible) MSY is usually >>MVP, but is <MVP in some local population.

Precautionary Approach? dN / dt = r [1 – (N / K) q ] N – fN, Maximize yield at f MSY = rq / (1 + q). If estimates of r, K, q and f includes uncertainty, MSY is not achieved, nor extinction risk is not eliminated. f < f MSY (Precautionary approach) is risk factor not to achieve MSY.

f MSY is neither sufficient nor necessary to stock conservation High uncertainty in rLow uncertainty in r

17 Conclusion: What is needed for PP Usually avoid Type II errors (risk-averse) Say a falsifiable prediction (responsibility of present assessment to the future) Show the weight of Evidence from unbiased estimates Non-regret policy (Acceptance of high risk from “good” manners)

Species Replacement of Pelagic Fishes Catch in Japan (1000 mt) Anchovy Horse mackerels Pacific saury Chub mackerel Sardine

Cyclic Advantage Hypothesis The next dominant to sardine is anchovy – Yes! As I predicted The second next is chub mackerel Many people agree now Matsuda et al. (1992) Res. Pop. Ecol. 34:

Future of Pelagic Fish Populations in the north-western Pacific: If overfishing of chub mackerel continues, –Chub mackerel will not recover forever; If cyclic replacement hypothesis is true, –Sardine will not recover forever; Do not catch immature mackerel too much –The overfishing is an experiment for my hypothesis. (Adaptive mismanagement)