• Gallagher, Austin J.
  • Lawrence, Michael J.
  • Jain-Schlaepfer, Sofia M. R.
  • Wilson, Alexander D. M.
  • Cooke, Steven J.


The nonconsumptive consequences of predators on prey behavior, survival, and demography have recently garnered significant attention by ecologists. However, the impacts of top predators on free-ranging prey are challenging to evaluate because the most common fright response for prey is to leave the area of risk. Additionally, the top-down impacts of avian predators on aquatic environments are surprisingly overlooked. Here we investigated the nonconsumptive effects of avian predators on parental care in pumpkinseed (Lepomis gibbosus (L., 1758)) through use of a realistic model of a predatory bird, the Osprey (Pandion haliaetus (L., 1758)). Our predator model exacted dramatic metabolic fright responses and inducible defenses in experimental fish resulting in significant behavioral changes with respect to their parental care. Key parental behaviors including in-nest rotations and egg and nest maintenance were noticeably altered by predator treatments demonstrating as much as an order of magnitude difference in parental performance, suggesting that even transient predation risk might decrease reproductive fitness. Our data provide important new insights on how the landscape of fear operates along the air–water interface and suggests that avian predators may have greater controlling effects on fish populations than previously thought.


Field observations occurred along a shallow gravel/sand bar (~500 m long x 100 m wide) containing a high abundance of pumpkinseed nests with males guarding eggs in the littoral zone of the central-western edge of the lake (an estimated 150 nests were constructed at the start of the study). All experiments were conducted between the hours of 1000 and 1400 under calm and sunny conditions. The depth of the sandbar ranged from 1 to 2 m, and the nesting substrate throughout this region was primarily composed of sand-gravel. For the purposes of this study, we focused on clean nests (i.e., excluding vegetation or rock-dominated nests) with males actively guarding eggs. Suitable nests were located via snorkeling by trained researchers (teams of two) and were marked with a white PVC identification tag (10 cm x 5 cm) on the outer rim. The total length (TL) of parental male pumpkinseed was estimated in situ by researchers, as precise measurements of fish size were not possible without capture and thereby resulting in potentially unacceptable levels of disturbance. Parental males were within the size range (estimated TL = 120-200 mm) of fish prey for Osprey (Hakkinen 1978; Francour and Thibault 1996); fish length classes were used (class 1 = small (120-140 mm); class 2 = medium (140-160 mm); class 3 = large (>160 mm)) and the relative size of each overall egg mass was scored by the same snorkeling team (classes 1-5, where class 1 = very few and class 5 = very many). All nests that we used contained an adult male and egg mass of unhatched, fertilized eggs (Colgan and Gross 1977). After each nest was selected and tagged, we then departed the nest area (i.e., 5-10 m away, outside the visual detection range of the fish) for a period of at least 90 s to mitigate disturbance to the fish before experimentation.

Each experimental series began with an aggression test that was conducted by one snorkeler, by placing a bluegill (Lepomis macrochirus Rafinesque, 1810) (an active nest predator; 100-140 mm total length) housed inside a clear glass container (volume ~4 L) on the edge of the nest for 60 s. The number of attacks (defined as any close rush, bump, or bite) made on the container by the parental male was recorded over the observation period, after which the jar was removed. This method is widely used to evaluate parental aggression in centrarchid fishes (Zolderdo et al. 2016). One snorkeler then placed a small high-definition camera on the edge of the nest (GoPro Hero 3+; Struthers et al. 2015) and recorded fish behavior continuously for a period of 12 min. This 12 min period consisted of three phases: (1) pre-stimulus, 5 min; (2) stimulus phase (presence/absence of predator model), 2 min; (3) post-stimulus, 5 min (Fig. 1). In phases 1and 3, we chose a set of parental behaviors scored as metrics following Cooke et al. (2008) and Colgan and Gross (1977) as represented by (i) cumulative time outside the nest (seconds), (ii) number of full (360°) in-nest rotations, (iii) total number of spinal flares (defined as the erection of dorsal spines), (iv) number of burst behaviors (defined as rapid, erratic swimming behaviors across the nest), and (v) total number of nest-maintenance behaviors (including any combination of tail fanning, mouthing of eggs, body-axis change to look at eggs). During phase 2, we quantified predation-risk-related behaviors (e.g., latency, defined as the mean duration from when the predator was introduced until the fish demonstrated a marked behavioral response, in seconds) and recorded the total number of trips each fish made outside the nest, as well as the number of burst swimming behaviors and dorsal spine erections exhibited by each individual. Across all trials, we also quantified the total number of individuals that temporarily left, as well as the number of individuals that temporarily abandoned and returned, versus those that never returned.