Indirect risk effects reduce feeding efficiency of ducks during spring

Abstract

Indirect risk effects of predators on prey behavior can have more of an impact on prey populations than direct consumptive effects. Predation risk can elicit more vigilance behavior in prey, reducing the amount of time available for other activities, such as foraging, which could potentially reduce foraging efficiency. Understanding the conditions associated with predation risk and the specific effects predation risk have on prey behavior is important because it has direct influences on the profitability of food items found under various conditions and states of the forager. The goals of this study were to assess how ducks perceived predation risk in various habitat types and how strongly perceived risk versus energetic demand affected foraging behavior. We manipulated food abundance in different wetland types in Illinois, USA to reduce confounding between food abundance and vegetation structure. We conducted focal-animal behavioral samples on five duck species in treatment and control plots and used generalized linear mixed-effects models to compare the effects of vegetation structure versus other factors on the intensity with which ducks fed and the duration of feeding stints. Mallards fed more intensively and, along with blue-winged teal, used longer feeding stints in open habitats, consistent with the hypothesis that limited visibility was perceived to have a greater predation risk than unlimited visibility. The species temporally nearest to nesting, wood ducks, were willing to take more risks for a greater food reward, consistent with an increase in a marginal value of energy as they approached nesting. Our results indicate that some duck species value energy differently based on the surrounding vegetation structure and density. Furthermore, increases in the marginal value of energy can be more influential than perceived risk in shaping foraging behavior patterns. Based on these findings, we conclude that the value of various food items is not solely determined by energy contained in the item but by conditions in which it is found and the state of the forager.

Keywords: foraging; nonlethal effects; perceived predation risk; risk‐taking; waterfowl.

Figure 1

Blue‐winged teal model predicted feeding intensity (probability of an observation to be classified as feeding) at various dates for each sex and treatment level (from model ‘date + sex + treatment’). Gray dotted lines represent predicted values plus or minus one standard error. Day 65 and 100 correspond with 6‐Mar and 10‐Apr, respectively

Figure 2

Model predicted feeding intensity for mallards at various water depths (topleft), flock sizes (topright), wetland types (bottomleft), and in treatment and control plots following treatment (bottomright). Error bars and gray dotted lines represent predicted values plus or minus one standard error. Water depth, flock size, and treatment * days since treatment plots were created from model ‘Water depth + Flock size + Treatment * Days since treatment’. The wetland type plot was created from the model ‘Water depth + Flock size + Wetland type + Treatment * Days since treatment’. In all plots, effects of interest were allowed to vary while other variables were held constant at their mean

Figure 3

Model predicted feeding intensity ± standard error for wood ducks at various water depths in treatment and control plots (topleft), flock sizes (topright), dates and sex (bottomleft), and days as treatment in treatment and control plots (bottomright). The water depth * treatment and flock size plot were created from model ‘Water depth * Treatment + Flock size’. The date * sex plot was created from model ‘Water depth * Treatment + Date * Sex’. The treatment * days since treatment plot was created from model ‘Water depth + Flock size + Treatment * Days since treatment’. In all plots, effects of interest were allowed to vary while other variables were held constant at their mean

Figure 4

Model predicted feeding intensity ± standard error for lesser scaup in different wetland types and treatments (from model “wet type + treat”)

Figure 5

Model predicted feeding intensity ± standard error for ring‐necked ducks different wetland types and treatments (left; from model “wet type + treat”) and at various days as treatment in treatment and control plots (right; from model ‘Wetland type + Treatment * Days since treatment, holding wetland type constant to open water)