Antipredator strategies of house finches: are urban habitats safe spots from predators even when humans are around?

Abstract

Urbanization decreases species diversity, but it increases the abundance of certain species with high tolerance to human activities. The safe-habitat hypothesis explains this pattern through a decrease in the abundance of native predators, which reduces predation risk in urban habitats. However, this hypothesis does not consider the potential negative effects of human-associated disturbance (e.g., pedestrians, dogs, cats). Our goal was to assess the degree of perceived predation risk in house finches (Carpodacus mexicanus) through field studies and semi-natural experiments in areas with different levels of urbanization using multiple indicators of risk (flock size, flight initiation distance, vigilance, and foraging behavior). Field studies showed that house finches in more urbanized habitats had a greater tendency to flock with an increase in population density and flushed at larger distances than in less urbanized habitats. In the semi-natural experiment, we found that individuals spent a greater proportion of time in the refuge patch and increased the instantaneous pecking rate in the more urbanized habitat with pedestrians probably to compensate for the lower amount of foraging time. Vigilance parameters were influenced in different ways depending on habitat type and distance to flock mates. Our results suggest that house finches may perceive highly urbanized habitats as more dangerous, despite the lower number of native predators. This could be due to the presence of human activities, which could increase risk or modify the ability to detect predators. House finches seem to adapt to the urban environment through different behavioral strategies that minimize risk.

Appendices

Appendix 1. Study areas used and degree of urbanization

The table presents the list of parks in more and less urbanized habitats and their sizes used for this study.

Table 2 List of parks in more and less urbanized habitats and their sizes

We quantified several aspects of these parks to confirm our classification. We obtained information on park area from the Long Beach Parks, Recreation and Marine website (http://www.longbeach.gov/park/facilities/parks) for parks located within Long Beach and from the County of Orange, Harbors, Beaches, and Parks website (http://www.ocparks.com/) for those located in Orange County.

Human disturbance was characterized in the parks where flock surveys were conducted and where FID measurements were taken using different methodologies. We measured the number of pedestrians in the 22 parks listed above by counting the number of people walking and sitting detected within the transects during the flock surveys. In addition, we measured pedestrian rate on the spots in which flight initiation distance had been recorded using two 50-m² squared plots to better characterize human disturbance in the samples areas: one plot was centered where the focal bird was before flushing, and the other plot was in a random location 50–100 m away from the previous plot. Both pedestrian rate measurements were taken on the same day, but days after we recorded FID. At the center of the 50-m² plot, during a 10-min period, we counted the number of pedestrians walking and sitting, as well as the number of dogs (leashed and unleashed) that entered the plot and averaged the values from both plots.

We measured vegetation structure in 25-m² plots at the center of every transect used to survey flocks. We visually estimated (following Prodon and Lebreton 1981): grass cover, bush cover, and tree cover. We measured mean bush and tree height by using a tape meter when physically possible. For taller trees, we estimated height by visually rotating the location of the tip of the tree onto the ground and then measuring the ground distance with a meter tape (±0.05 m; following Fernández-Juricic et al. 2005). We also recorded the number of tree trunks >30-cm dbh), as these represented potential perches for house finches (Fernández-Juricic et al. 2005).

Appendix 2. Description of the experimental procedures to assess scanning and foraging behavior

The semi-natural experiment was carried out from September to November 2005 in two locations separated by 3.87 km: Seal Beach (33° 45′ N, 118° 05′ W; less urbanized) and CSULB campus (33° 46′ N, 118° 06′ W; more urbanized).

Forty-two wild adult house finches were caught and color-banded from two Southern California populations: Seal Beach (33° 45′ N, 118° 05′ W) and Irvine (33° 39′ N, 117° 51′ W). We decided to capture birds from areas that had intermediate degrees of urbanization to avoid habituation effects towards humans. House finches were housed in 0.85 × 0.6 × 0.55-m cages under a 12:12-h light–dark cycle (lights went on 06:00) and were kept in visual and auditory contact with four birds per cage (male and female house finches were housed in the same cages). No breeding behavior was observed during the experimental period. Water and food (finch seed mix, Royal Feeds, Leach Grain and Milling, Co., Downey, CA, USA) were provided ad libitum except during experimental trials and the preceding periods of food deprivation. At the end of the experiments, birds were released at the same locations in which they were captured.

Our experimental setup consisted of three sets of two bottomless circular light wire mesh enclosures (0.5-m diameter, mesh opening 0.06 m, 85% open area) placed in a parallel arrangement. Each enclosure was placed on a round wooden tray and connected by a link (0.3 × 0.18 × 0.18 m) to the other enclosure of the same set (Fig. 4). One of the enclosures (hereafter, foraging patch) contained 3 cm of sawdust and 5.00 ± 0.05 g of hidden sunflower seeds. The other enclosure (hereafter, refuge patch) had synthetic foliage placed around the top of the cage and one synthetic bush with the base in the center of the refuge. We placed one bird in each set of enclosures during the experiment, totaling three birds per trial. We considered the individual in the central set as the focal animal to minimize edge effects.

Fig. 4

Semi-natural experimental design testing habitat type and neighbor distance. Each enclosure consists of a foraging patch connected by a corridor to a refuge patch. This example shows the 2 m (far) neighbor distance treatment

We manipulated two different treatments: habitat type (three levels: less urbanized without pedestrians, more urbanized with pedestrians, and more urbanized without pedestrians) and neighbor distance (two levels: close, 0-m separation between the sets of enclosures, and far, 2-m separation between the sets of enclosures, Fig. 4). Thus, each focal bird was exposed to six different treatments (one per day) in a random order. Trials were performed in the mornings from 07:00 to 12:00 but not during high winds or rain.

The treatments in the more urbanized habitat were conducted at CSULB campus on a grassy area 20 m from a pathway frequented by pedestrians and 15 m from a large old pine tree and a four-story campus building. For the more urbanized habitat without pedestrian treatment, we surrounded the experimental area with a 2.0-m-high fence and black tarp to screen out visual stimuli (e.g., pedestrians), but the building was still in the visual field of animals. The purpose of this treatment was to obtain the baseline behavior of birds without the influence of pedestrians or habitat openness that would trigger changes in vigilance to monitor for a potential predator. The downside is that it created a visual obstruction effect that may have blocked the need for vigilance.

For the more urbanized habitat with pedestrians treatment, we used the same location without the fence so that birds were exposed to pedestrians as well as an assistant walking in a parallel path 9 m away from the enclosures to simulate the presence of humans. The less urbanized habitat treatment was performed at Seal Beach in an area without a fence. This experimental site was not near any human structures or trees and did not have any pedestrian traffic. The vegetation consisted of native bushy plants and bare ground. The differences in habitat structure between these sites mirrored the ones found between more and less urbanized parks.

Out of the 42 birds, 14 served as focals (seven males and seven females) and were used in a total of 84 trials. The remaining 28 non-focal birds (conspecifics) were assigned to a focal bird and went through all trials with the same focal individuals but were randomly assigned to the right and left sets of enclosures. All birds were food-deprived for 2 h prior to the experiments and transported to the CSULB experimental site by walking and to the Seal Beach location by car. Animals were then placed in the shade until their trial time. During transport, and, while awaiting their trial, they were provided water ad libitum.

The use of a car as a means of transporting the animals to the less urbanized habitat to run the semi-natural experiment could have produced changes in their behavior. To test this possibility, we ran another experiment at only the CSULB campus site, in which house finches experienced two treatments. In the first treatment (driving), birds were driven for 10 min and then placed back in their aviary cages for 30 min before they were used in the experiments. This treatment was conducted with the same car used during the experiments and replicated the less urbanized treatment procedures in the main experiment. In the second treatment (non-driving), birds were not driven at all but were walked to the experimental site replicating the more urbanized treatment procedures in the main experiment. The same three sets of enclosures were placed 0.5 m apart (Fig. 4), with one bird in each under similar conditions as described in the “Materials and methods.” We screened out visual stimuli by fencing around the cages in all treatments. We recorded the foraging and scanning behavior of eight focal birds during 10 min, with each focal being exposed to each treatment. The birds used in this experiment were different from those used in the one presented in the main text.

None of the secondary factors (temperature vs. food deprivation time, r = 0.24, P = 0.255; temperature vs. body condition, r = 0.30, P = 0.150; food deprivation time vs. body condition, r = 0.04, P = 0.840) was correlated among each other. Using general linear models (repeated-measures design), we found that the proportion of time spent in the refuge did not vary between treatments (driving, 0.03 ± 0.01; non-driving, 0.02 ± 0.01; F _{1, 4} = 1.39, P = 0.303), controlling for temperature (F _{1, 4} = 0.00, P = 0.968), food deprivation time (F _{1, 4} = 0.07, P = 0.811), and body condition (F _{1, 4} = 2.82, P = 0.168). Scan rate (driving, 18.19 ± 2.25 events per minute; non-driving, 16.83 ± 2.25 events per minute; F _{1, 4} = 0.38, P = 0.572) and scan bout duration (driving, 2.25 ± 0.29 s; non-driving, 2.83 ± 0.29 s; F _{1, 4} = 2.12, P = 0.219) did not differ between treatments, controlling for the three secondary factors (scan rate: temperature, F _{1, 4} = 1.51, P = 0.287; food deprivation time, F _{1, 4} = 1.26, P = 0.324; body condition, F _{1, 4} = 0.11, P = 0.754; scan bout duration: temperature, F _{1, 4} = 5.60, P = 0.077; food deprivation time, F _{1, 4} = 3.15, P = 0.151; body condition, F _{1, 4} = 1.14, P = 0.346). Finally, there was no difference in pecking rates between treatments (driving, 21.75 ± 2.86 events per minute; non-driving, 18.28 ± 2.86 events per minute; F _{1, 4} = 1.29, P = 0.321), controlling for the secondary factors (temperature, F _{1, 4} = 6.28, P = 0.066; food deprivation time, F _{1, 4} = 5.16, P = 0.0.86; body condition, F _{1, 4} = 0.09, P = 0.776). Therefore, we conclude that driving the birds to conduct the less urbanized treatment would not substantially affect their behavior.

During the main experiment, the focal individual was recorded for 20 min using a Sony DCR-TRV38 digital video camera starting from the time it began pecking in the foraging patch. The camera was placed on a tripod 3 m from the focal’s set of enclosures, which allowed us to record its scanning and foraging behavior while in the foraging patch and its use of the refuge patch. We recorded behavior using an event-recording program (JWatcher 1.0; Blumstein et al. 2006). When birds were in the foraging patch, they could be on the ground or hanging on the cage wall. We only focused on their behavior on the ground as a proxy of normal foraging behavior. House finches alternated bouts of food-seeking behavior with their heads down and bouts of scanning behavior with their heads up. We recorded the number of head-up scanning events and their duration, including head-up food-handling behavior as they use it for vigilance (Fernández-Juricic and Tran 2007). While house finches were head-down, we recorded the number and duration of pecking events. We calculated the following dependent variables: head-up scan bout duration, head-up scan rate, and instantaneous pecking rate (number of pecking events per unit time head-down). We also recorded and calculated the proportion of time the focal bird spent in the refuge patch. Video analysis was performed by AV after extensive self-training. At the time of recording, there was less than a 5% difference between two scorings of the same tape. For every trial, we recorded several potential confounding factors: food deprivation time, ambient temperature, wind speed, and focal body condition (body mass/wing length).