Community structure may vary spatially and temporally. A number of univariate indices have been developed to quantify community structure, although much information is lost in this approach, which results in a single number to describe an entire community. Multivariate methods provide more information-rich results, although they may be more difficult to interpret. Nonmetric multidimensional scaling (NMS) is a nonparametric ordination approach that reduces the dimensionality of a complex data set to two or three dimensions, which can be easily interpreted visually. Communities can be compared in reduced-dimensionality ordination space, and important habitat or environmental variables may be inferred by evaluating the strength of the correlation between each variable and the ordination axes. We used NMS as part of a long-term monitoring project to compare structure in fish communities in two river systems: the Buffalo National River (BUFF) in northern Arkansas, and the Current River and its major tributary the Jacks Fork at Ozark National Scenic Riverways (OZAR) in southern Missouri. Goals were to compare community structure both among sites and over time, and elucidate potentially important habitat variables. Six sites were selected at BUFF, and nine at OZAR, and sampled annually for five years by electrofishing.
NMS ordinations revealed that sites at OZAR were generally well separated in ordination space and were characterized by very similar community structure over time. In other words, most sites were distinctly different from each other, regardless of the year sampled. In contrast, at BUFF the position of sites in ordination space varied greatly over time; communities often grouped together by year sampled more strongly than by site. In other words, communities changed from one year to the next, exhibiting patterns of nonrandom annual variability. With a single exception (watershed area), the habitat variables most strongly correlated with ordination axes differed between the two river systems. The rivers at OZAR receive substantial flow from numerous large springs, and thus base flows tend to be more constant in terms of temperature and volume. The river at BUFF, in contrast, is fed primarily by numerous small tributaries, and varies more dramatically in these characteristics over time. Water temperature, for example, was strongly correlated with the ordination axes at OZAR, but not at BUFF. This study demonstrates the usefulness of the NMS ordination approach in ecological monitoring, and reveals how spatiotemporal variation in fish communities may vary among different types of river systems.