一种改进的多学习教与学优化算法

针对传统教与学算法在解决复杂多峰函数优化问题时,具有局部最优且搜索开发能力较差的缺点,提出了一种改进的多学习教与学优化算法,新算法为学员的每一维加入不同的教学因子,设计了基于学员均值比较的教师选择策略和向教师及学员学习的多学习策略。基于多个单峰和多峰函数的仿真结果表明,新算法跟传统的、改进的教与学算法相比,在稳定性、寻优精度和收敛速度方面更具优势。     

Effects of specimen handling and otolith preparation on concentration of elements in fish otoliths

The use of otolith chemistry as a tool for analysing the biology of fishes assumes that the procedures used to collect and prepare otoliths for analysis do not alter their composition. With otoliths of Nemadactylus macropterus, Hoplostethus atlanticus, and Rhombosolea tapirina, we show that this assumption is not valid for the elements that can be detected using electron-probe microanalysers (those present at concentrations greater than ≃100 parts per million): all six elements routinely measured using these techniques were affected by at least one post-mortem procedure tested. Measured concentrations of calcium and strontium were relatively insensitive to most procedures tested, whereas concentrations of sodium, potassium, sulphur and chlorine were affected substantially by many commonly used procedures. The ease with which otolith composition could be modified post-mortem suggests that apparent geographic, habitat-specific or ontogenetic differences in otolith composition should be interpreted with extreme caution because of easily induced artefacts and the problem of pseudo-replication associated with the ways otoliths are sampled and prepared for analysis. Received: 23 September 1996 / Accepted: 11 April 1998     

Framework for participative reflection on the accomplishment of transdisciplinary research programs

In response to the increasingly complex social–ecological issues facing society, there is a growing trend to conduct environmental research in large collaborative programs. This approach is described as transdisciplinary research as it transcends formal disciplinary boundaries, explicitly acknowledges that many different perspectives are relevant to the resolution of complex problems, and actively involves the users of research. This poses challenges for the evaluation of “impact” as any evaluation process must take into consideration the different expectations, values, culture, language and reward structures of the main participating groups, the funders, researchers and end users. How can these participating groups learn about the progress of a transdisciplinary research program in a way that is purposeful and structured, continues through the life of the program, and includes explicit feedback mechanisms that facilitate adaptation during the course of the program? This paper presents a framework for co-reflecting on the accomplishment of transdisciplinary research programs. The framework incorporates the perspectives of funders, researchers and users, and recognizes that while they place different emphasis on measures of achievement such as efficiency, rigor and relevance, ultimate accomplishment in terms of translating knowledge into practice requires that the needs and expectations of all three groups are adequately addressed. What emerges from the framework is the importance of early investment in processes, behaviors and relationships that foster social learning and the co-production of the knowledge and understanding that are required to ensure relevance; while maintaining emphasis in the traditional areas of formally testing evidence and mentoring young researchers to ensure rigor and build confidence and capacity in transdisciplinary approaches.     

Genetic divergence and phylogenetic inferences in five species of Mugilidae (Pisces: Perciformes)

Allozyme electrophoresis was used to compare the genetic divergence of global populations of Mugil cephalus Linnaeus, 1758 and two congeneric [M.␣curema Cuvier & Valenciennes, 1836; M. gyrans (Jordan & Gilbert, 1885)] and two more distantly related [Liza ramada (Risso, 1826); Xenomugil thoburni (Jordan & Starks, 1896)] species on the basis of 18 enzyme loci. The amount of genetic divergence among the species examined is in agreement with their present systematic status, the differences being larger among non-congeneric species than among species of the same genus. Intraspecific genetic distances in M. cephalus (average Nei's D= 0.154), although higher than those reported for conspecific populations of fish, appear to be small when compared to the interspecific values among mugilid species (0.821 ≤ Nei's D ≤ 1.744). Phylogenetic trees obtained by genetic distance methods and discrete character parsimony analysis were of similar topology, except for the relationships within the genus Mugil and for the arrangement of M. cephalus populations. Received: 7 April 1997 / Accepted: 4 February 1998