Background/Question/Methods Understanding how communities respond to environmental change is frustrated by the fact that both movement and species interactions affect biodiversity in unseen ways. Dynamic joint species distribution models (JSDM) that include movement are needed to evaluate the contributions of species interactions and movement. We used citizen science data through eBird to evaluate these impacts using a dynamic JSDM that captures both types of interactions. A novel aspect of our model allows for redistribution across a spatial region. We adopt a Bayesian approach, enabling probabilistic uncertainty quantification in the model parameters. Additionally, we incorporate methods to handle the issue of zero inflation via regressions. Recognizing uneven sampling effort, a challenge often associated with leveraging opportunistic data is accounting for biases in data collection. Motivated by the citizen science eBird data, we first develop a model for sampling bias in spatial locations and effort. Sampling is a point process with preferential sampling that is accommodated through a shared-process model. Our method to scale point-level effort to areal-unit level yields a regressor in our JSDM. Results/Conclusions Our effort model reveals how environmental covariates influence where birders sample and how much effort they put into an observation. We find preferential sampling, in that the amount of effort at a location is not independent of the location itself. We also demonstrate how to accommodate misalignment between point-referenced data and areal units. For our dynamic JSDM, simulations and analytical analysis demonstrate the ability to capture interactions between species and with the environment. Applications to a selection of resident and migratory birds from eBird data examine if and how species-environment interactions impact abundances for our choice of species. Results show that the strength of a species' interactions with its environment and with other species can be a dominant effect on individual growth rates. We show that poorly-defined movement and an abundance of observed zeros degrade estimates of species interactions. Through its dynamic and probabilistic framework, our model provides predictions of abundance. Implications for conservation planning include showing environmental conditions where species interactions are strong, and vice versa. Additionally, we contribute to methods that enable the use of citizen science data for ecological analyses.