The power of the Bayesian model to detect the response of different functional groups
The individual hypothesis for colonization and permanence of functional groups using the space-state Bayesian model yielded at least three different response patterns: i) the estimated parameters of permanence and colonization confirm the hypothesis raised; ii) at least one of the parameters confirm our hypothesis; or iii) no parameter fit the hypothesis.
The Chironomidae and the Ceratopogonidae larvae are clear examples of the first case. Their probabilities of permanence and colonization were very close to the hypothesized probabilities. During the experiment we found a low probability of permanence, but a high and moderate probability of colonization for Chironomidae and Ceratopogonidae, respectively. This indicates that these organisms invest primarily in a rapid re-colonization of habitats, which is in line with the life cycle of these organisms. The representatives of these families usually have high densities and a relatively short life cycle (Trivinho-Strixino and Strixino 1999). The life span of Ceratopogonidade and Chironomidae females can reach 70 and 22 days, respectively (Braverman 1994), period during which the females have two to four reproductive cycles (Corbi and Trivinho-Strixino 2006).
Only the parameters estimated for the group composed by Coleoptera larvae with depressed bodies (e.g., Macrelmis , Phanocerus ,Xenelmis ) did not support our hypotheses. Contrary to our expectation, these larvae had a low permanence during the study period accompanied by a moderate habitat colonization capacity. An explanation for this pattern may be the mobility capacity of these larvae, and their active search for habitats (Carvalho and Uieda 2006) that allows an alternated occurrence of these organisms in the samples.
Our hypotheses were supported by at least one of the estimated parameters of the remaining six groups. Many explanations may account for these results, and we need highlight that some parameters may have been overestimated in some hypotheses because we had limited previous information about the biology of some groups. Despite this Hutchinsonian shortfall, mobility of Coleoptera larvae, other uses to shelters (e.g., Trichoptera using the shelter to protect against possible natural enemies, Morse 2009), rapid colonization in empty habitat (Ephemeroptera and Plecoptera, Carvalho and Uieda 2004, 2006, Landeiro et al. 2010), and the use of drift strategies for colonization (Megaloptera and some Coleoptera and Trichoptera larvae, Krueger and Cook 1984, Townsend and Hildrew 1994, Godoy et al. 2016) may explain the divergence between the hypothesis and patterns observed in this study.
A relevant point in our study was the way we used the functional groups. The groups were assembled according to life strategies based on functional traits. In addition, we established testable numerical hypotheses for predetermined parameters. Both biology of organisms and ecological theory were included in a predictive way (Verberk et al. 2013). The elaboration of these hypotheses and our experimental design in space-state made it possible for us to measure the permanence and colonization responses separately. The separate estimation of the parameters that represent these functional responses was essential to understand how changes in flow conditions can impact the occurrence of organisms. We also directly tested the axes of resistance and resilience that summarize the fundamental theoretical functional traits of aquatic insects undergoing disturbances in the environment (Lopez et al. 2016). Separately, functional groups respond differently to changes in the flow regimes, even for groups with similar baseline permanence and colonization values.
Understanding the response of functional groups to different disturbances is an important tool to elaborate theories and hypotheses that explain temporal dynamics of aquatic communities, and to describe the possible alternative states presented by communities depending on environmental stress. Despite the separate estimates of probability of permanence and colonization, the experimental design does not allow the total separation between these two parameters and, thus, the estimated parameters may have some level of covariance. The long-term permanence probability may be the result of a continuous occupation of organisms, partially affected by colonization processes. Despite this limitation, the use of a space-state analysis reduces this covariance to acceptable limits. To control and quantify this covariance, a similar study limiting the habitat colonization is necessary. It is also noteworthy that our experimental design was based on a categorical quantification of the disturbances, leaving out a possible continuum for flow disturbance intensity and frequency. However, this qualification is common in studies that assess the effect of the disturbance on biodiversity (Poff and Zimmerman 2010) and the levels we choose have relevance to the type of environment we test our hypotheses.