Abstract. Nearly all ecosystems have been altered by human activities, and most communities are now composed of interacting species that have not co-evolved. These changes may modify species interactions, energy and material flows, and food-web stability. Although structural changes to ecosystems have been widely reported, few studies have linked such changes to dynamic food-web attributes and patterns of energy flow. Moreover, there have been few tests of food-web stability theory in highly disturbed and intensely managed freshwater ecosystems. Such synthetic approaches are needed for predicting the future trajectory of ecosystems, including how they may respond to natural or anthropogenic perturbations.We constructed flow food webs at six locations along a 386-km segment of the Colorado River in Grand Canyon (Arizona, USA) for three years. We characterized food-web structure and production, trophic basis of production, energy efficiencies, and interaction-strength distributions across a spatial gradient of perturbation (i.e., distance from Glen Canyon Dam), as well as before and after an experimental flood. We found strong longitudinal patterns in food-web characteristics that strongly correlated with the spatial position of large tributaries. Above tributaries, food webs were dominated by nonnative New Zealand mudsnails (62% of production) and nonnative rainbow trout (100% of fish production). The simple structure of these food webs led to few dominant energy pathways (diatoms to few invertebrate taxa to rainbow trout), large energy inefficiencies (i.e., ,20% of invertebrate production consumed by fishes), and right-skewed interaction-strength distributions, consistent with theoretical instability.Below large tributaries, invertebrate production declined ;18-fold, while fish production remained similar to upstream sites and comprised predominately native taxa (80-100% of production). Sites below large tributaries had increasingly reticulate and detritus-based food webs with a higher prevalence of omnivory, as well as interaction strength distributions more typical of theoretically stable food webs (i.e., nearly twofold higher proportion of weak interactions). Consistent with theory, downstream food webs were less responsive to the experimental flood than sites closest to the dam. We show how human-induced shifts to foodweb structure can affect energy flow and interaction strengths, and we show that these changes have consequences for food-web function and response to perturbations.

While the ecology and evolution of partial migratory systems (defined broadly to include skip spawning) have been well studied, we are only beginning to understand how partial migratory populations are responding to ongoing environmental change. Environmental change can lead to differences in the fitness of residents and migrants, which could eventually lead to changes in the frequency of the strategies in the overall population. Here, we address questions concerning the life history of the endangered Gila cypha (humpback chub) in the regulated Colorado River and the unregulated tributary and primary spawning area, the Little Colorado River. We develop eight multistate models for the population based on three movement hypotheses, in which states are defined in terms of fish size classes and river locations. We fit these models to mark–recapture data collected in 2009–2012. We compare survival and growth estimates between the Colorado River and Little Colorado River and calculate abundances for all size classes. The best model supports the hypotheses that larger adults spawn more frequently than smaller adults, that there are residents in the spawning grounds, and that juveniles move out of the Little Colorado River in large numbers during the monsoon season (July–September). Monthly survival rates for G. cypha in the Colorado River are higher than in the Little Colorado River in all size classes; however, growth is slower. While the hypothetical life histories of life-long residents in the Little Colorado River and partial migrants spending most of its time in the Colorado River are very different, they lead to roughly similar fitness expectations when we used expected number of spawns as a proxy. However, more research is needed because our study period covers a period of years when conditions in the Colorado River for G. cypha are likely to have been better than has been typical over the last few decades.

Dams and river regulation greatly alter the downstream environment for gross primary production (GPP) because of changes in water clarity, flow, and temperature regimes. We estimated reach-scale GPP in five locations of the regulated Colorado River in Grand Canyon using an open channel model of dissolved oxygen. Benthic GPP dominates in Grand Canyon due to fast transport times and low pelagic algal biomass. In one location, we used a 738 days time series of GPP to identify the relative contribution of different physical controls of GPP. We developed both linear and semimechanistic time series models that account for unmeasured temporal covariance due to factors such as algal biomass dynamics. GPP varied from 0 g O 2 m 22 d 21 to 3.0 g O 2 m 22 d 21 with a relatively low annual average of 0.8 g O 2 m 22 d 21 . Semimechanistic models fit the data better than linear models and demonstrated that variation in turbidity primarily controlled GPP. Lower solar insolation during winter and from cloud cover lowered GPP much further. Hydropeaking lowered GPP but only during turbid conditions. Using the best model and parameter values, the model accurately predicted seasonal estimates of GPP at 3 of 4 upriver sites and outperformed the linear model at all sites; discrepancies were likely from higher algal biomass at upstream sites. This modeling approach can predict how changes in physical controls will affect relative rates of GPP throughout the 385 km segment of the Colorado River in Grand Canyon and can be easily applied to other streams and rivers.