Summary For six fish species sampled from 86 lakes on the Arctic Coastal Plain, Alaska, we examined whether lake occupancy was related to variables representing lake size, colonisation potential and/or the presence of overwintering habitat. We found the relative importance of each factor for a given species could be related to its ecology and adult size. The three large‐bodied migratory species, least cisco (Coregonus sardinella), broad whitefish (Coregonus nasus) and arctic grayling (Thymallus arcticus), were influenced by factors associated with the likelihood of fish recolonising lakes, including whether the lakes had a stream connection. Of the large‐bodied species, least cisco had the highest likelihood of occupancy (0.52 ± 0.05) and models provided evidence that least cisco exhibit both migratory and resident forms. Models for small‐bodied fish differed among species, indicating different niches. Ninespine stickleback (Pungitius pungitius) were the most widespread and ubiquitous of the species captured (occupancy probability = 0.97 ± 0.01); they were captured in lakes that freeze to the bottom, suggesting that they disperse widely and rapidly after the spring freshet, including colonisation of sink habitats. Alaska blackfish (Dallia pectoralis) had a lower occupancy (occupancy probability = 0.76 ± 0.05) with a distribution that reflected tolerance to harsh conditions. Slimy sculpin (Cottus cognatus) had an occupancy probability of 0.23 ± 0.06, with a distribution indicating its marine origin. Based on these patterns, we propose an overall model of primary controls on the distribution of fish on the Arctic Coastal Plain of Alaska. Harsh conditions, including lake freezing, limit occupancy in winter through extinction events while lake occupancy in spring and summer is driven by directional migration (large‐bodied species) and undirected dispersal (small‐bodied species).

Interspecific competition is an important process structuring ecological communities, however, it is difficult to observe in nature. We used an occupancy modelling approach to evaluate evidence of competition between yellow-billed (Gavia adamsii) and Pacific (G. pacifica) loons for nesting lakes on the Arctic Coastal Plain of Alaska. With multiple years of data and survey platforms, we estimated dynamic occupancy states (e.g. rates of colonization or extinction from individual lakes) and controlled for detection differences among aircraft platforms and ground survey crews. Results indicated that yellow-billed loons were strong competitors and negatively influenced the occupancy of Pacific loons by excluding them from potential breeding lakes. Pacific loon occupancy was conditional on the presence of yellowbilled loons, with Pacific loons having almost a tenfold decrease in occupancy probability when yellow-billed loons were present and a threefold decrease in colonization probability when yellow-billed loons were present in the current or previous year. Yellow-billed and Pacific loons co-occurred less than expected by chance except on very large lakes or lakes with convoluted shorelines; variables which may decrease the cost of maintaining a territory in the presence of the other species. These results imply the existence of interspecific competition between yellow-billed and Pacific loons for nesting lakes; however, habitat characteristics which facilitate visual and spatial separation of territories can reduce competitive interactions and promote species co-occurrence.

Studies examining species occurrence often fail to account for false absences in field sampling. We investigate detection probabilities of five gear types for six fish species in a sample of lakes on the North Slope, Alaska. We used an occupancy modeling approach to provide estimates of detection probabilities for each method. Variation in gear- and species-specific detection probability was considerable. For example, detection probabilities for the fyke net ranged from 0.82 (SE = 0.05) for least cisco (Coregonus sardinella) to 0.04 (SE = 0.01) for slimy sculpin (Cottus cognatus). Detection probabilities were also affected by site-specific variables such as depth of the lake, year, day of sampling, and lake connection to a stream. With the exception of the dip net and shore minnow traps, each gear type provided the highest detection probability of at least one species. Results suggest that a multimethod approach may be most effective when attempting to sample the entire fish community of Arctic lakes. Detection probability estimates will be useful for designing optimal fish sampling and monitoring protocols in Arctic lakes.

1. Surface water connectivity can influence the richness and composition of fish assemblages, particularly in harsh environments where colonisation factors and access to seasonal refugia are required for species persistence. 2. Studies regarding influence of connectivity on Arctic fish distributions are limited and are rarely applied to whole assemblage patterns. To increase our understanding of how surface water connectivity and related hydrologic variables influence assemblage patterns, we investigated species richness and composition of Arctic lake fishes over a large region, 8500 km 2 , of the central Arctic Coastal Plain, Alaska. 3. We collected fish presence/non-detection data from 102 lakes and used a hierarchical multispecies occupancy framework to derive species richness and inform species composition patterns. Our mean estimate of regional richness was 12.3 (SD 0.5) species. Presence of a permanent channel connection was an overriding factor affecting species richness (mean 3.6, 95% CI 3.1-4.9), presumably driving lake colonisation potential. In lakes without a permanent channel connection, data suggest richness (mean 2.0, 95% CI 1.7-3.3) increased with the availability of in-lake winter refugia and with the potential of ephemeral connections during spring floods. 4. Fish species functional traits and environmental faunal filters contributed to patterns of richness and assemblage composition. Composition corresponded with richness in a coherent manner, where each successive level of richness contained several discrete assemblages that showed similar responses to the environment. Lakes with permanent channel connections contained both widespread and restricted species, while the species-poor lakes that lacked a connection contained mainly widespread species. 5. This work provides useful baseline information on the processes that drive the relations between patch connectivity and fish species richness and assemblage composition. The environmental processes that organise fish assemblages in Arctic lakes are likely to change in a warming climate.