measures of assemblage similarity, there is a greater risk of losing such species either from local habitats or system-wide. Because of this, although responses of assemblages to perturbations indicate generally high stability, postimpact fish assemblages (even if judged highly similar to preimpact assemblages by most measures of assemblage structure) might differ in the loss of rare species.
Although there are fewer lentic compared to lotic studies, those in lentic systems tend to compare longer time intervals (median 15 versus 11 years) and half as many species (mean 10, range 1–20, versus mean 21, range 3–95). Compared to lotic studies, lentic studies were also generally at higher latitudes (mean 49, range 34–64, versus mean 36, range 31–42)—regions that typically have lower fish species diversity. Thus our understanding of persistence and stability in streams, reservoirs, and lakes is incomplete because of relatively few studies and biases in geographic location, species richness, and length of comparisons.
Persistence and Stability of Local Associations
The previous sections dealt with levels of change in species assemblages over time periods of two or more years and over moderate to broad spatial scales. Much less is known about how close contacts of species in associations change over time—for instance how long do multispecies groups remain and do they remain together long enough for reciprocal evolutionary responses (coevolution) to occur? It can be difficult to detect association patterns in species of mobile animals. Individuals found in the same sample may or may not have been in close enough contact to have had direct interactions with each other. The capture of individuals of two species in a sample may not equate to their direct interaction because most survey methods for fishes, or other mobile vertebrates, have fuzzy boundaries and may sample different microhabitats (Ross and Matthews, in press). Also, fishes found together in a relatively long reach of stream (e.g., Marsh-Matthews and Matthews 2002) may, especially in highly structured habitats, occupy different pools or riffles and thus do not encounter each other daily. For mobile animals distributed across a heterogeneous landscape, consistent spatial associations among species could exist because some species (or some life stages within a species) select similar microhabitats totally independent of each other (see also Chapter 13) (Chapman and Chapman 1993; Grossman et al. 1998; Wilson 1999).
TABLE 6.2 Long-Term (≥ 2 years) Studies of North American Lake and Reservoir Fish Assemblages Organized from Low to High Levels of Perceived Stress and from Low to High Latitudes
For a downloadable PDF of all tables, go to ucpress.edu/go/northamericanfishes
Matthews and Marsh-Matthews (2006a) provided one of the clearest studies of the longevity of multispecies associations. Based on 19 snorkeling surveys taken over 22 years, they examined persistence of associations of eight taxonomic species (including minnows, a topminnow, sunfishes, and black basses) and 11 “ecospecies” (with the sunfishes separated into piscivorous adults versus insectivorous juveniles) across 14 adjacent pools within a kilometer reach of Brier Creek, Oklahoma (Figure 6.6). For each completed survey of the 14 pools (Figure 6.6A), species associations were compared by constructing a triangular similarity matrix of species pairs based on relative abundances (Figure 6.6B). Next, the strength of species associations over time was determined by sequentially comparing the 18 matrices using the Mantel test, a statistical procedure for comparing the correlation between matrices (Legendre and Legendre 1998). Concordance (based on Z-scores provided by the Mantel test) declined as the interval between samples increased (Figure 6.6D), so that associations within a year were largely concordant, but associations across years within a season were concordant only in late summer. Overall, species associations were concordant for approximately half of the 18 intervals between snorkeling surveys. Associations were not typically changed by events like floods and droughts, but the second of two very severe droughts in three years coincided with distinct changes in associations of species or ecospecies. The empirical evidence of Matthews and Marsh-Matthews (2006a) suggests that although some smaller species groups (pairs, triads, or foursomes) might remain consistently in direct contact across years or even decades, there was little evidence that whole assemblage, multispecies associations were constant, or that selection pressures due to such multispecies groups would be consistent across such periods of time.
Persistence, Stability, and Control of Fish Assemblages
As pointed out at the beginning of the chapter, an important reason for trying to understand the levels of persistence and stability in fish assemblages is that assemblages that show high persistence, and particularly those with high stability, may be more influenced by biotic interactions (and have greater likelihood of deterministic control) than by abiotic factors (with greater likelihood of stochastic control). Clearly, the level of physical disturbance can influence the position of a community along a gradient of deterministic to stochastic control, as suggested by various authors, with biotic interactions likely more important in communities with low levels of disturbance and abiotic factors more important in communities with high levels of disturbance (e.g., Grossman et al. 1982; Peckarsky 1983). This is not to say that all assemblages showing stability are deterministically controlled or that all assemblages lacking stability are stochastically controlled. For instance, an assemblage with strong deterministic control based on competitive interactions among species could show a lack of stability because of differential time lags in the effects of species interactions (Strong 1983).
FIGURE 6.6. Persistence of associations among fish taxa in Brier Creek, Oklahoma, based on 19 snorkeling surveys taken over 22 years.
A. A simple example of similarity matrices based on two surveys with three taxa and three pools, where sxy is the similarity of taxa x and y between two pools.
B. Actual data showing the strength of the associations between consecutive samples relative to the time between the surveys; the solid line is the regression line based on the array of Z-scores. Based on Matthews and Marsh-Matthews (2006a).
One way of addressing the question of the degree of deterministic control of fish assemblages would be to follow an assemblage over time in the absence of major disturbance. However, such systems are uncommon in nature, except for some isolated springs. An alternative approach would be to use a seminatural, artificial stream system. If all streams offer essentially the same environments, then deterministic control should result in high similarity among fish assemblages. Matthews and Marsh-Matthews (2006b) stocked seven outdoor artificial streams with identical numbers and kinds of species and then followed assemblage composition for 388 days. Even though the streams were as identical as possible, differences did develop over time in the extent of algal cover and in the level of predation. Different levels of predation were caused by the differential survival of sunfishes among pools. Somewhat surprisingly, even in the absence of natural disturbances, the assemblages diverged significantly in composition, so that the ultimate structure of any of the experimental assemblages “could not be predicted from its initial structure.” In other words, the study did not support predictions of strong deterministic control.
SUMMARY
Fish assemblages change over both ecological and evolutionary time scales. Assemblages controlled primarily by random processes (i.e., primacy of stochastic control) have greater variation in species composition and abundances, compared to those influenced primarily by nonrandom processes (i.e., primacy of deterministic control). Disturbances include any event that disrupts a community or a particular assemblage in some way. Measures of assemblage responses to disturbance include those of persistence (i.e., the presence or absence) of fish species or by stability, a measure that includes the kinds of species and their abundances. Responses
