ABSTRACT In this paper, our goal was to test fish–environment relationships at headwater streams using three different taxonomic levels (species, family, and order) considering two hypotheses: i) fish–environment relationships are expected to occur at higher taxonomic levels; ii) fish–environment relationships are expected to differ among distinct families and orders. We sampled 37 headwater streams using the electrofishing method and we collected information for nine environmental variables: channel depth and width, luminosity, pH, dissolved oxygen, water temperature, conductivity, turbidity, and velocity. For each level, we constructed one occurrence matrix and one environmental matrix to test their co-structure through a Co-inertia analysis followed by a Monte-Carlo test. We found 75 species belonging to six fish orders and 23 families. Our findings revealed significant fish–environment relationships at headwater streams for both species and family levels but not for order level. We found that the set of variables affecting species occurrence is also responsible to influence the family’s distribution. These results may indicate that, in the lack of information for species, the family level be used in substitution to seek for fish–environment relationships. However, before that, more studies are needed in order to show whether these patterns are consistent for other regions of the world.

中文翻译：

---

环境对较高鱼类分类水平的影响：热带源头溪流的关系

摘要 在本文中，我们的目标是使用三个不同的分类级别（物种、科和目）来测试上游河流中的鱼类 - 环境关系，考虑两个假设：i）鱼类 - 环境关系预计发生在更高的分类级别；ii) 不同科和目之间的鱼类-环境关系预计会有所不同。我们使用电捕鱼法对 37 条源头水流进行了采样，并收集了九个环境变量的信息：通道深度和宽度、光度、pH 值、溶解氧、水温、电导率、浊度和速度。对于每一层，我们构建了一个发生矩阵和一个环境矩阵，通过共惯性分析和蒙特卡罗测试来测试它们的共结构。我们发现了属于 6 个鱼类目和 23 个科的 75 个物种。我们的研究结果揭示了在物种和科水平的源头溪流中显着的鱼类 - 环境关系，但不是为了秩序水平。我们发现影响物种发生的变量集也负责影响家庭的分布。这些结果可能表明，在缺乏物种信息的情况下，可以使用科级别来替代寻找鱼类与环境的关系。然而，在此之前，还需要更多的研究来证明这些模式是否与世界其他地区一致。在缺乏物种信息的情况下，可以使用家庭级别代替寻找鱼类与环境的关系。然而，在此之前，还需要更多的研究来证明这些模式是否与世界其他地区一致。在缺乏物种信息的情况下，可以使用家庭级别代替寻找鱼类与环境的关系。然而，在此之前，还需要更多的研究来证明这些模式是否与世界其他地区一致。