1. Conservation and management increasingly focus on connectivity, because connectivity driven by variation in immigration rates across landscapes is thought to be crucial for maintaining local population and metapopulation persistence. Yet, efforts to quantify the relative role of immigration on population growth across the entire range of species and over time have been lacking.

2. We assessed whether immigration limited local and range-wide population growth of the endangered snail kite (Rostrhamus sociabilis) in Florida, USA, over 18 years using multi-state, reverse-time modeling that accounts for imperfect detection of individuals and unobservable states. Demographic contributions of immigration varied depending on the dynamics and geographic position of the local populations, were scale-dependent, and changed over time.

3. By comparing the relative contributions of immigration versus local demography for periods of significant change in local abundance, we found empirical evidence for a disproportionately large role of immigration in facilitating population growth of a centrally-located population—a connectivity ‘hub’. The importance of connectivity changed depending of the spatial scale considered, such that immigration was more important driver of population growth at small versus large spatial scales. Furthermore, the contribution of immigration was much greater during time periods when local population size was small, emphasizing abundance-dependent rescue effects.

4. Our findings suggest that efforts aimed at improving local breeding habitat will likely be most effective at increasing snail kite population growth. More broadly, our results provide much needed information on the role of connectivity for population growth, suggesting that connectivity conservation may have the greatest benefits when efforts focus on centrally located habitat patches and small populations. Furthermore, our results highlight that connectivity is highly dynamic over time and that interpreting the effects of connectivity at local scales may not transfer to region-wide dynamics.

We used mark-resight data across the breeding range to interpret the role of immigration on population dynamics. From 1997 through 2015, multiple standardized band-resight surveys (4-6) were conducted during the peak of the snail kite breeding season (March 1 –June 30^th) covering the entire known breeding range of the snail kite in Florida (Dreitz et al., 2002; Martin, Kitchens, & Hines, 2007; Reichert, Kendall, et al., 2016). During this time, 3,327 individually banded snail kites were resighted (7,232 total resights, of which 4,522 were in the northern region, 798 in Lake Okeechobee, and 1912 in the southern region). Of the total resighted individuals, 1,289 were banded before 1997, and first observed within this study as adults (>1 year of age). From 1997 to 2015, 2,888 snail kites were banded in the nest just prior to fledging (1,361 in the northern region, 252 on Lake Okeechobee, and 1,275 in the southern region) and 45 were banded as adults (28 in the northern region, 3 in Lake Okeechobee, and 14 in the southern region).

This is a multi-state reverse-time robust band-resight datset for snail kites covering their entire breeding range in Florida from 2015-1997. "0" = non-detected,  "S" = snail kites resighted as adults (>1 year old) in the southern sub-population (south of Lake Okeechobee),  "R" = snail kites captured and banded as young (<1 year old) in the southern sub-population (south of Lake Okeechobee), "K" = snail kites resighted as adults (>1 year old) in Lake Okeechobee,  "J" = snail kites captured and banded as young (<1 year old) in Lake Okeechobee, "N" = snail kites resighted as adults (>1 year old) in the northern sub-population (north of Lake Okeechobee),  "M" = snail kites captured and banded as young (<1 year old) in the northern sub-population (north of Lake Okeechobee). 

All years have 6 repeat survey efforts, except 2002 and 2003, which have 5 and 4, respectively. 

To assess the role of immigration in maintaining local population growth, we compared the relative contribution of total immigration (adult and juvenile) versus the combined contributions of local adult survival (and fidelity) and local juvenile recruitment for each of the three local populations (northern region, southern region, and Lake Okeechobee).  We tested whether contribution parameters varied over time using AIC to compare the relative fit of a model with time-invariant contribution parameters to that of a model that allowed contributions to vary among years. All models were implemented using Program MARK 8.1 (White & Burnham, 1999), implemented using the R-package, RMark (Laake, 2013).