Abstract: [Objective]Organization and market of forestry industry cluster has essentially ecological and social of groups, but the spillover effects of "industrial air" on each forestry enterprise which produce innovation performance are different. In this study, 82 forestry enterprises in forestry industry cluster of Fujian Province were taken as research objects, to hackle the mechanism of how cluster network structure generates innovation performance and quantitatively distinguish key factors that cause the differences of enterprise innovation performance. The conclusions are to provide references for formulating innovation policies of forestry industry cluster and achieving the transformation and upgrading of forestry industry.[Method]Firstly, based on the theory of corporate social capital and the characteristics of the forestry industrial cluster, the innovation network structure for enterprises of forestry industry cluster was constructed. The influence mechanism of network structure and network relationship on the enterprise innovation performance was analyzed. Then the social network analysis research paradigm which was widely used by domestic and foreign scholars was adopted to depict the characteristics of innovation activities for enterprises of forestry industry cluster from the network structure and the network relationship, and the research hypothesis was also put forwards. Secondly, using Likert five point table method to design measurement table, the dependent variable (i.e. innovation performance) and the independent variables, including the network structure characteristic variables (i.e. network scale, network density, network openness, network intermediary, network resources richness) and the network relationship characteristic variables (i.e. network relationship stringency, network relationship reciprocity, network relationship stability, network feedback mechanism) were determined. At the same time, the reliability and validity of the measurement table was also test. Then, based on the data of 82 forestry enterprises collected by questionnaire survey method, and analyzed by SPSS16.0 software to take OLS fitting for the regression model, the results showed that regression model present multicollinearity problems, so it was necessary to amend the OLS regression model. Finally, using R software to apply partial least squares regression and construct the model, the key factors which impact on innovation performance differences among enterprises of forestry industry cluster were obtained.[Result]The results showed that:1) From the perspective of innovation network structure of forestry industry cluster, network intermediary (1% significant level) and network resources richness (1% significant level) had significantly positive influences on the enterprise innovation performance. 2) From the perspective of network relationships, network feedback mechanism (5% significance level) also had a significantly positive effects on the enterprise innovation performance, while network reciprocity (5% significance level) had significantly negative effects on the enterprise innovation performance.[Conclusion]With the acceleration of transformation and promotion of forestry industry, forestry enterprise innovation has been increasingly depended on the network environment, the superior network structure of forestry industry cluster innovation would be an indispensable condition for enterprises to carry out innovation activities, especially the intermediate position of enterprises in the cluster network and the rich innovation resources in cluster, including knowledge, information and talent, etc., which will enhance the ability of forestry enterprises in cluster to dominate innovation resources. In addition, good relationship among cluster innovation network is a fine booster for successful innovation evolution. Although the conclusion in this study that network reciprocity has a significantly negative influence on the enterprise innovation performance do not consistent with other industries' results, however, it in a certain extent fit with the development stage of Fujian forestry industry cluster and the characteristics of forestry industry chain. This paper finally gives some advice on how to improve the innovation performance of forestry industry cluster by the enterprise and government aspects, respectively.

Key words: forestry industrial cluster, network structure, innovation performance, partial least-squares