田中 嘉成

生物多様性と生態系機能の関係性の解明と、それに基づいた生態系の影響評価のための新たなアプローチとして、生物の機能形質（生態形質）に基づいた枠組みが進展している。理論的な面では、群集レベルでの機能形質の動態や、生態系機能の応答の定式化のために、集団遺伝学や量的遺伝学の進化理論が応用されている。実証データと理論的枠組みの連携がさらに進めば、生物の分布情報、環境要因データ、生態形質のデータベースに対する統合的な解析から、生態系影響評価が可能になると期待される。

In this paper, both the empirical and theoretical genetic aspects of human-mediated introgressive hybridization are reviewed in terms of their association with the breakdown of postzygotic isolating mechanisms. I also compare several simulation models with an ecological or genetic focus that are relevant to the prediction and risk assessment of genetic extinction due to hybridization. One barrier to devising comprehensive risk assessment frameworks is a lack of sufficient population genetic studies that associate introgressive hybridization with specific isolating mechanisms. A gametic model based on multilocus underdominant fitness is one of the best genetic models for introgressive hybridization because it explicitly incorporates the postzygotic isolating mechanism known as Dobzhansky-Muller genetic incompatibility.

Achievements of the research project entitled "Establishment of a scientific framework for the management of toxicity of chemicals based on environmental risk-benefit analysis" supported by the JST were introduced and reviewed, focusing on the development of the methodology for estimating risks; human health risks and ecological risks. The usefulness of loss of life expectancy as a metric for evaluating cancer and noncancer risks was demonstrated. To evaluate ecological risks, three metrics, 1/T, logT and T, developed based on the mean extinction time (T) of species were proposed. Then, their implication and feasibility were examined in terms of what ecological system should be conserved and how easily people can understand the implications of metrics. Protocols for estimating human health risks and ecological risks are illustrated. (C) 2003 Elsevier Ltd. All rights reserved.

Studies on ecological risk assessment achieved by the department of theoretical ecology in recent years are briefly reviewed especially on the ecotoxicological experiments, the meta-analysis on published ecotoxicological data and the analytical method to estimate extinction risk of pollutant chemicals. The life table evaluation data, which estimated toxicant stress on the intrinsic rate of natural increase, were reviewed and analyzed with specific concentration-response functions. Among analyzed mathematical models (the power function model, the Weibull mode, and the quadratic function model), the power function model exhibited the best fit to observed concentration-response curves. Based on the best-fit response function, extinction risks by several important pollutant chemicals were estimated for Daphnia poptllations with the mathematical solution of mean extinction time as a function of the intrinsic rate of natural increase. The estimated extinction risks differed quantitatively from the traditional ecological hazard quotients, which dismissed population-level effects.

Theories in population ecology and genetics are reviewed in the context of their application to biological conservation. Specific branches of the whole topic are mean extinction time models based on diffusion equations, extinction risk due to onbreeding depression and chemical pollutants, adaptation under environmental changes, and hetapopulation dynamics. The importance of unification between population ecological and population genetical approaches in stressed for resolving these problems.

For the ecological risk assessment, extrapolation from toxicological data obtained at the individual level into effects at the population level is required. We review the analytical methods for translating chronic toxicity data into the effect on propensity of populations (population growth rate or intrinsic rate of population increase). Actual ecotoxicological data have two major problems, i.e., 1) only a very small fraction of chemicals-species combinations has been examined for chronic toxicity, and 2) it is not feasible for many test species to execute the life-cycle test. As analytical methods in order to circumvent these limitations of data, we focus on the extrapolation method and the life history sensitivity analysis, and discuss these methods in the context of ecological risk assessment. The cxtrapolation method is to infer missing chronic toxicity data from regression of known chronic data to acute data, or from regression of chromic data between different species or life stages. From the inferred and the directly estimated chronic toxicity data, the effect of chemicals to population growth rate is estimated. The life history sensitivity analysis estimates the relative importance of life stages in terms of intrinsic rate of natural increase, and reduces the life table evaluation by excluding the unimportant life stages. These analytical methods that apply the ecological theory may be important for future ecotoxicological data analysis embedded in the ecological risk assessment.

Ecotoxicological data using life table evaluation are reviewed and analyzed with the power function model. Life table evaluation and population growth experiments are proposed as experimental procedures that provide demographic parameters (e.g. intrinsic rate of natural increase) relevant for calculating extinction risk due to pollutant exposure. Totally 47 concentration vs. intrinsic rate data sets were collected and analyzed with two indices in the power function model, α and β. The α-value is the concentration at which the intrinsic rate of increase drops off below zero due to exposure and the β-value represents curvature of the response. The α-values, which represent strength of ecological toxicity, are highly correlated with acute LC50s. It is indicated that the α-values are indirectly predictable from acute LC50s. Such statistical extrapolation may be useful for ecological risk assessment based on extinction probability of populations.

A framework of ecological risk assessment based on extinction probability of populations is presented. The estimation of extinction risk consists of three steps, i.e. 1) estimation of dose-intrinsic natural increase curve based on experimental evaluation of dose-response curves, 2) calculation of dose-extinction time curves by means of analytical solutions of extinction time models, and 3) from comparison between the dose-extinction time curves and real exposure of pollutants into natural environments, the extinction risk of target species is evaluated. In this paper, I reviewed previous ecological toxicity experiments, and concluded that the life table evaluation is the most stringiest protocol for estimating important population parameters ( especially, intrinsic rate of natural increase). To summarize the data base of life table evaluation I analyzed the data with a power function model for dose-intrinsic rate curves. Based on this model, available data can be summarized into two summary parameters, i.e. α and β, which respectively represents the critical dosage where the intrinsic rate drops off below 0 and the curvature of the curves. There is a highly positive correlation between α-value and LD50s, suggesting a possibility of extrapolation from acute toxicities to α-values. Extinction time models are shortly reviewed, and extinction cost was tentatively calculated using these models.