Abstract
In the Galapagos, humans have established a permanent presence and have altered selective pressures through influences such as invasive predators and urbanization. I quantified flight initiation distance (FID), an antipredator behavior, in Darwin’s finches, across multiple islands in the Galapagos to ask: (i) does antipredator behavior (e.g. FID) change in the presence of invasive predators and importantly, what happens once they have been eradicated and (ii) to what degree does urbanization affect antipredator behavior? This is one of the first studies to quantify behavior in an endemic species after successful eradication of invasive predators. FID was higher on islands with invasive predators compared to islands with no predators. On islands from which invasive predators were eradicated ∼11 years previously, FID was also higher than on islands with no invasive predators. Within islands that had both urban and non-urban populations of finches, FID was lower in urban finch populations, but only above a threshold human population size. FID in larger urban areas on islands with invasive predators was similar to or lower than FID on islands with no history of invasive predators. Overall, these results suggest that invasive predators can have a lasting effect on antipredator behavior, even after eradication, and that the effect of urbanization can strongly oppose the effect of invasive predators, reducing antipredator behavior to levels lower than found on pristine islands with no human influences. These results improve our understanding of human influences on antipredator behavior which can help inform future conservation and management efforts on islands.
Significance Statement Humans are reshaping the selective pressures that organisms experience through urbanization and the introduction of invasive predators. Invasive predators have been successfully eradicated on islands, for example in the Galapagos, but we lack a comprehensive understanding of how endemic species will cope after such eradication. Furthermore, we do not fully understand how much urbanization can counter antipredator adaptations. I found that Darwin’s finches maintain increased antipredator in response to introduced predators even after successful eradication, and that urbanization can strongly counter such behavioral adaptations by greatly reducing antipredator behavior in the presence of invasive predators. Human influences can have strong, counteracting effects on endemic species and their behavioral traits, and these must be considered in island conservation and management efforts.
“All of [the terrestrial birds] are often approached sufficiently near to be killed with a switch, and sometimes, as I myself tried, with a cap or a hat.” – Charles Darwin in “The Voyage of the Beagle”
Introduction
Human influences such as invasive species and urbanization can strongly affect the process of local adaptation (1–4). Such effects are amplified on islands such as the Galapagos Islands, where small population sizes and strong isolation increase the vulnerability of local flora and fauna to human influences, often resulting in loss of island biodiversity through extinctions (5– 8). Among the endemic species on the Galapagos Islands are Darwin’s finches, an iconic example of an adaptive radiation in which a single founding species has evolved into several species, each with different adaptions (e.g. beak shapes and body sizes) to exploit different ecological niches (9, 10). Humans began establishing settlements on the Galapagos in the early 19th century (11), and since then, human influences such as invasive predators and urbanization have affected several islands on the Galapagos. Many organisms initially respond to such human influences through behavioral adaptations. Here, I consider how two human influences – invasive predators and urbanization – might alter antipredator behavior in Darwin’s finches on the Galapagos Islands.
Invasive predators have strong ecological and evolutionary effects (1, 2, 12, 13), and this impact is known to be correlated with local extinction events (14, 15). On islands, the lack of predators and correlated relaxed selection can result in reduced antipredator behavior (16–19). This evolutionary naïveté of isolated animals that have evolved without major predators can contribute to the extirpation of island species (17, 18, 20, 21). In particular, feral and domestic house cats (Felis silvestris catus) are of concern for island biodiversity because cats target small animals such as birds and reptiles (22–25), and invasive house cats now exist on four islands of the Galapagos (26), presenting a critical threat for Galapagos biodiversity (27, 28). Past research on the effects of invasive predators in the Galapagos has focused on behavioral adaptations in reptiles (e.g., 27, 29), and thus, little is known about the effect of novel mammalian predators on endemic land birds. Given the resulting selective pressures, natural selection should favor an increase in antipredator behavior after the introduction of an invasive predator to reduce mortality.
Effective conservation management, especially on islands, often involve eradication of invasive predators to protect the local and endemic species (30, 31). Post-eradication research typically follows local and endemic species population recovery (32–34), monitors the re-introduction of extirpated species to previously abandoned breeding grounds (35), or focus on major ecological effects such as changes in food web dynamics (32, 36). All this research contributes to the growing need to understand post-eradication effects (33, 34, 37), yet surprisingly little research has focused on post-eradication behavioral adaptations, nor how quickly such behavioral adaptations might occur. Post-eradication behavioral adaptations could have population-level consequences on fitness. For example, increased antipredator behavior can have associated costs due to the reallocation of energy and time away from other important behaviors such as foraging, reproduction, and rearing of young (38–40). Thus, if increased antipredator behavior is maintained after eradication, this might result in a decrease in fitness for local and endemic species. Understanding how local and endemic species will behaviorally adapt post-eradication could help improve conservation efforts. On the Galapagos, some islands have invasive house cats, some have remained free of invasive predators, and some islands have successfully eradicated invasive predators. This allows for among-island comparisons of antipredator behavior in relation to the current and historical invasive-predator regime.
Urbanization has rapidly increased in the past century, with more than half the world’s population occupying urban settlements, severely altering patterns of selection and adaptation (3, 4, 41). In general, animals such as birds show decreased antipredator behavior in urban areas compared to rural areas, likely due to habituation to humans (42–45). However, such a reduction in antipredator behavior in urban areas could make organisms more vulnerable to different threats, such as invasive predators. Quantifying the degree to which urbanization can reduce antipredator behaviors can inform our understanding of the impacts of urbanization. Can urbanization reduce antipredator behavior to levels before the introduction of predators?
The Galapagos Islands represent an excellent opportunity to study the effects on invasive predators and urbanization for two key reasons that few, if any, other systems offer. First, few other archipelagos in the world have islands that vary not only in the presence or absence of invasive predators, but also have islands that have successfully eradicated invasive predators. Second, islands differ not only in the presence or absence of urban centers, but also in the size of the urban centers, representing a novel opportunity to compare antipredator behavior along a gradient of urbanization as well as among islands that differ in the presence or absence of urban centers. The isolation of the Galapagos Islands removes potentially confounding factors such as high gene flow or continued influxes of introduced predators, allowing me to ask two key questions regarding human influences and antipredator behavior. First, I ask how will antipredator behavior change in the presence of invasive predators and perhaps more importantly, what will happen after eradication of invasive predators from an island? Very little research has been done on behavioral adaptations post-eradication. Second, I ask how much can urbanization reduce antipredator behavior – can it be reduced to levels found on islands with no history of invasive predators? While we know how urbanization can affect behavior, we have no sense of the degree to which urbanization is affecting behavior because it is difficult to find a system in which we can assess baseline behavior before urbanization occurred (e.g. islands with no history of permanent human populations). Together, these two questions can inform how human influences are affecting antipredator behavior on isolated islands.
Results
How does antipredator behavior change in the presence of invasive predators and what happens following eradication of invasive predators from an island?
On islands with invasive predators and islands where predators have been eradicated, FID in finches was significantly higher than on pristine islands (Table 1, Figure 1). Post-hoc comparisons showed that finches on pristine islands (Santa Fe and Española) had lower FID when compared to finches on islands with invasive predators (Figure 1; p = 0.054) and eradicated islands (Figure 1; p = 0.012), and finches on eradicated islands did not differ in FID when compared to finches on islands with invasive predators (Figure 1, p = 0.179). As group size increased, FID increased (Table 1, Supplemental Figure 1). Analysis with data only for small ground finches had comparable results with the only differences being group size was no longer significantly correlated with FID (Table 1) and a post-hoc comparison of FID from finches on islands with invasive predators was significantly lower than finches on pristine islands (p = 0.015). When considering the effect of island independently of invasive-predator regime, island and group size had a significant effect on FID (Figure 1; Table 2). Post-hoc comparisons of FID within an invasive-predator regime showed no significant difference in FID between finches on the two pristine islands (p = 0.677), among the four islands with invasive predators (0.105 < p < 0.591) with the exception of finches on Floreana having significantly higher FID than finches on Santa Cruz (p = 0.009). Between the two eradicated islands, finches on Baltra had significantly higher FID than finches on North Seymour (p = 0.010). Analysis with data only for small ground finches had comparable results with the only differences being group size was no longer significantly correlated with FID (Table 2) and a post-hoc comparison of FID of islands within a predation regime found finches on Baltra did not significantly differ from finches on North Seymour (p = 0.996) and finches on Floreana did not differ significantly from finches on Santa Cruz (p = 0.532).
How much can urbanization affect antipredator behavior?
Urban finches had significantly lower FID as compared to non-urban finches (Table 3, Figs. 1 and 2). A post-hoc linear mixed model analyzing FID in finches on Floreana showed no significant difference between urban and non-urban populations (χ2 = 0.358, p = 0.550) whereas FID in finches was significantly lower in urban areas as compared to non-urban areas on Isabela (χ2 = 19.062, p < 0.001), San Cristobal (χ2 = 28.478, p < 0.001), and Santa Cruz (χ2 = 22.099, p < 0.001). On San Cristobal, finches in urban populations had lower FID than found on pristine islands (Figure 1). Group size was positively correlated with FID (Table 3, Supplemental Figure 1), and time of day was also positively correlated with FID (the later in the day it was, the higher the FID; Table 3, Supplemental Figure 2). Analysis with data only for small ground finches had comparable results (Table 3) with the only differences being island and time of day no longer had a significant effect on FID (Table 3).
Discussion
Increased antipredator behavior is maintained after eradication of invasive predators
Finches exhibited increased antipredator behavior on islands with invasive predators (Table 1; Figure 1). More interestingly, this increased antipredator behavior was also observed on islands where invasive predators had been eradicated (in 2003 and 2008, 13 and 8 years prior to data collection; the mean generation time of finches is one year, by comparison). Since all islands have naturally occurring local or endemic predators of finches (Supplemental Table 1), this effect was most likely due to the presence of invasive predators, even after eradication. This is one of the first studies to show that increased antipredator behavior has been maintained on islands that have had invasive predators removed. Several possible reasons exist for these observations, especially the apparent maintenance of elevated FID on eradicated islands. First, it is possible increased antipredator behavior has evolved on islands that have and used to have invasive predators. However, without knowledge of heritability, and thus actual evolution, this cannot be confirmed, but would be an area for future research. Second, perhaps the expected costs of increased antipredator behavior are not high enough to cause a reversion to pre-predator levels. Increased FID can have associated costs (38–40), suggesting that if this behavioral adaptation were costly then finches on eradicated islands would have FID comparable to finches on pristine islands. It could also be that not enough time has elapsed for reversion in antipredator behavior. Third cultural transmission of increased FID (62, 63) with learned behavior transmitted from generation to generation could maintain the increased FID. Lastly, the increase in FID could be due to something other than predation, which could still be present on eradicated islands, or that for unknown historical reasons, antipredator behavior on the eradicated islands have historically been high. The last reason is possible, but it would be a quite a coincidence if the eradicated islands that have this elevated FID for some reason unrelated to predation just happen to be exactly the islands where predation was introduced and then eradicated.
Regardless of the mechanism, the fact that antipredator behavior levels did not revert post-eradication (when comparing FID on eradicated islands to FID on pristine islands; Figure 1) has potential consequences for evaluating the efficacy of eradication efforts. Recent studies of local animal populations post-eradication have focused on demographic parameters such as population recovery (32–34) or on ecological parameters such as food-web dynamics (32, 36). However, such phenomena will be influenced by behavioral shifts. For example, increased antipredator behavior correlates with decreased time and energy for behaviors such as foraging, courting, defending territories, or caring for offspring (38–40), which could affect population recovery and/or food-web dynamics. Thus, understanding how the eradication of invasive predators will affect the behavior of local or endemic animals should be central to future conservation efforts (33, 34, 37).
Urbanization can decrease antipredator behavior to levels lower than before the introduction of predators
Finches in urban areas had lower FID than finches in non-urban areas, supporting previous findings (42–45). However, two interesting points are found in this general trend. First, the degree of urbanization appears to determine just how much lower FID is for urban finches when compared to non-urban finches. Finches in the town of Puerto Velasco Ibarra on Floreana had the highest FID compared to finches in other towns, and Puerto Velasco Ibarra is also the smallest town, with a permanent population of only 111 (Supplemental Table 1). The significantly lower FID of finches in larger towns as compared to non-urban finches suggests an urbanization “threshold”, such that the degree of urbanization needs to be high enough to exert sufficient selective pressure on finches to drive behavioral adaptation; in short, perhaps Puerto Velasco Ibarra is simply too small to be an “urban” site for the purposes of finch behavioral adaptation. The next largest town, Puerto Villamil on Isabela, with a population of 2,164 (Supplemental Table 1), had significantly lower FID than Puerto Velasco Ibarra on Floreana. Puerto Villamil is still a relatively small town (Supplemental Table 1), showing that the threshold amount of urbanization sufficient to produce differences in antipredator behavior is not very high.
The second interesting point is that on some islands, urbanization can result in FID that is lower than FID on islands that have never been exposed to predators (Figure 1), even though urban areas invariably contain invasive predators such as cats and rats. In other words, in some towns, such as Puerto Baquizo Moreno on San Cristobal, FID has been reduced to levels below what was observed on islands with no history of invasive predators. Such reductions in FID in urban finches is likely due to habituation (64–67), suggesting that habituation from urbanization is so strong that it results in FID lower than the baseline FID quantified on pristine islands, counteracting any increase in FID due to the presence of invasive predators. This suggests that the effects of urbanization on organisms can be quite strong, with likely evolutionary and ecological consequences (3, 4, 68).
Group size and species
For all analyses with all species, FID significantly increased with increasing finch group size (Supplemental Figure 1). This supports the “many-eyes hypothesis” that detection of predators occurs earlier in large groups, when the predator is further away, due to the larger number of individuals watching (38). Because of the unbalanced design with respect to sample sizes of different species on different islands, and because Darwin’s finches are closely related species (implying statistical non-independence among species), I repeated all analyses with a subset of the data utilizing FID from the one species, the small ground finch, Geospiza fuliginosa. Species did not have a significant effect on FID in any of the main analyses, and these additional single-species results did not alter the interpretation of any of my results. Future work should use a more balanced sample size and to allow adjustment for non-independence among species.
Conclusions
Our current understanding of how humans affect the evolution and ecology of Darwin’s finches has primarily focused on beak shape evolution and shifts in ecological niches in response to changes in food availability and diet (69–71). However, humans have had further effects on the environment of the Galapagos through the creation of urban environments and the introduction of invasive mammalian predators. The Galapagos Islands represent an opportune system to study the effects of human influences due to the among island differences in invasive-predator regime as well as the differences in the amount of urbanization. Such systems do not readily exist elsewhere. Here, I showed how antipredator behavior increased in response to invasive predators and was maintained even after eradication of those invasive predators, and this can have possible demographic and ecological effects. This is one of the first studies to look at post-eradication behaviors in local and endemic species, which can have implications for future conservation efforts. I also found that urbanization can reduce antipredator behavior to levels at or below what was found on pristine islands, attesting to the strength of the effect urbanization can have on behavioral traits. Understanding the effects of different human influences will help us predict how organisms might respond to their rapidly changing environments.
Materials and methods
Site descriptions
The Galapagos Islands are a volcanic archipelago located ∼1,000 km off the coast of Ecuador. Local or endemic predators such as owls or snakes are found on all islands (Supplemental Table 1; Swash & Still 2005). Snakes are thought to prey on the nestlings of ground finches, and are thus an unlikely predator. However, short-eared owls, found on all islands surveyed in this study, are known predators of adult land birds such as Darwin’s finches (47). Galapagos Hawks, also predators of ground finches (48, 49), are found on four (Santa Fe, Española, Isabela, and Santa Cruz) of the eight islands surveyed in this study (46, 50). Unfortunately, little data are available about the current densities of local and endemic predators on these islands; however, the ecology of the local and endemic predators (e.g. owls) is well documented, and can thus be assumed to be predators of finches. The invasive-predator regime (presence of house cats) on the islands (Supplemental Table 1) were classified as: present (Floreana, Isabela, San Cristobal, Santa Cruz), pristine (Santa Fe, Española), or successful eradication (Baltra, North Seymour). House cats and rodents were successfully eradicated from Baltra in 2003 (26, 51), and rats from North Seymour in 2008 (52, 53); I will refer to these as “eradicated” below for brevity. The two pristine islands and two eradicated islands have no permanent human populations. On the four islands with human populations (Floreana, Isabela, San Cristobal, Santa Cruz), site urbanization categories were classified as: urban (in town) or non-urban (remote areas several kilometers away from town and not visited by tourists). Islands with permanent human populations and no presence of invasive predators nor pristine islands with permanent human populations exist in the Galapagos archipelago, and thus, I am restricted to the among island comparisons outlined above.
Data collection
Flight initiation distance (FID), the distance at which a prey flees an approaching predator, is a metric used to quantify antipredator behavior (38–40). An individual’s decision to flee is influenced by the perceived costs and benefits of remaining or taking flight, which means FID can be an indicator of how an organism assesses risk, and thus, antipredator behavior. Data were collected from 2015 to 2018 on eight islands of the Galapagos archipelago, generally between February and April (some data on San Cristobal were collected in November 2017).
FID measurements were performed with a human stimulus following methods from Blumstein (54). A focal finch was located by walking and searching the landscape at a slow walking pace, and the finch’s initial behavior was noted. To minimize the possibility of pseudoreplication, in a given day, each trial ensured the focal finch was of a different sex, species, or age class (for male Geospiza spp.) than finches that had previously been approached. However, it is possible the same bird might have been approached on different days or years because the finches were not individually banded. Birds were located in areas that had relatively open habitat to ensure a straight approach by the human and a clear sightline from the human to the finch. The human would then approach the focal finch at a standardized speed (∼2.2 m/s). Human stimuli always wore neutral-colored clothing, and looked at the focal individual while approaching.
Flight was considered to have been initiated if the finch extended its wings and flew; the distance flown could be short (<0.5 m) or substantial (out of sight). Finches that hopped away instead of taking flight were omitted from the study, though this was a rare occurrence. A marker was placed where the finch originally was and where the stimulus was when the finch took flight, and the distance between these markers was the FID. Because of the complexity of the landscape, the distance at which the stimulus started from could not be standardized (54–56), and so I noted the distance from where the human started to the flight-initiation marker (starting distance). Alert distance, the distance at which an individual is aware of the approaching stimulus (57), could not be quantified because the focal individual was often foraging on the ground and would repeatedly raise its head, and so normal foraging behavior was indistinguishable from an alert reaction to an approaching human.
Each data point collected included the island, invasive-predator regime, urbanization category (on islands with permanent human populations), finch species and sex, time of day, and group size (was the focal finch in a group and if so, how large was the group). Finch sex was identified by plumage, and for ambiguous cases and non-sexually dimorphic species (e.g. warbler finches), sex was denoted as unknown. Time of day was noted because birds are most active at dawn and at dusk, so baseline activity levels and behaviors can vary throughout the day. Group size was noted because it could increase FID because larger groups mean more observers and thus, detection of a potential threat will occur when the threat is still a longer distance away (38). Conversely, group size could decrease FID through the dilution effect where the probability a predator will target a specific individual decreases as group size increases (40). Island size was also noted (58) to account for potential among island environmental differences due to different area.
Statistical analysis
All analyses were done in R (version 3.4.3). To meet assumptions of normal distributions, FID and starting distance (the distance between the focal finch and the stimulus starting position) were log-transformed. Then, FID and starting distance were centered by subtracting the mean from each value and then dividing by the standard deviation. Because starting distance could not be standardized and is known to affect FID, it was included as a covariate, as was group size, time of day, and island size. Lastly, sex was included as a fixed effect and species and island was included as a random effect. Sample sizes for some species across islands were unbalanced (Supplemental Table 2); therefore, I also repeated all analyses using one species, Geospiza fuliginosa, the small ground finch. Further analysis details are below, but in general, linear mixed models were performed using lmer() from the lme4 package (59) and Anova() from the car package (60). Random-effect significance (species and/or island) was determined with ranova(), and post-hoc pairwise comparisons for fixed factors in the linear mixed models used difflsmeans(), both from the lmerTest package (61). R2 values were calculated with the r.squaredGLMM() from the MuMIn package (version 1.42.1).
How does antipredator change in the presence of invasive predators and what happens following eradication of invasive predators from an island?
On islands that had both urban and non-urban populations, only data from non-urban sites were used for this analysis. A linear mixed model was performed with invasive-predator regime and sex as a fixed factor, island and species as random factors, and time of day, starting distance, island size, and group size as covariates; post-hoc comparisons focused on pairwise comparisons between invasive-predator regimes. This analysis was performed twice, once with all finch species, and once with only small ground finches. The interaction between invasive-predator regime and sex was non-significant for both analyses and was removed.
To determine whether FID varied among islands within a given invasive-predator regime, I ran a linear mixed model with island and sex as fixed factors, species as a random factor, and time of day, starting distance, and group size as covariates; post-hoc comparisons were focused on pairwise comparisons between islands within a given invasive-predator regime. This analysis was repeated as a linear model with only small ground finches. The interaction between island and sex was non-significant for both analyses and was removed. A Tukey’s HSD (honestly significant difference) test was used for post-hoc comparisons for this analysis.
How much can urbanization affect antipredator behavior?
Only data collected from the four islands with permanent populations were used for this analysis. A linear mixed model was performed with site urbanization and sex as fixed factors, island and species as random factors, and time of day, starting distance, island size, and group size as covariates. This analysis was performed twice, once with all finch species, and once with only small ground finches. The interaction between invasive-predator regime and sex was non-significant for both analyses and was removed. To look at within island differences in FID among urban and non-urban finches, for each island, a linear mixed model was performed with the same factors as above without island as a random factor.
Acknowledgements
I wish to thank Marc-Olivier Beausoleil, Lara Bernasconi, Carlos Camacho, Sofia Carvajal, Angela Hansen, Célina Leuba, and Ashley Saulsberry for their assistance in the field. Permits were obtained with the assistance of Sofia Carvajal, Jaime Chaves, Jeff Podos, the Charles Darwin Research Centre, the Galápagos Science Centre, the Galápagos National Park, and the Universidad de San Francisco. I thank Sarah Knutie for numerous helpful suggestions on ideas and methods, Ben Haller, Andrew Hendry, and Claire Spottiswoode for extensive comments and friendly reviews of the manuscript, and Tom Houslay for guidance with statistics. My research has been generously supported by the Natural Sciences and Engineering Research Council of Canada (Banting Postdoctoral Fellowship), Le Fonds Québécois de la Recherche sur la Nature et les Technologies (postdoctoral research fellowship), Clare Hall (Whitehead Fund), Christ’s College (Galápagos Islands Visiting Scholarship Scheme), Newnham College (Phyllis and Eileen Gibbs Travelling Research Fellowship), the British Ornithologists’ Union, and McGill University (Department of Biology Vineberg Fellowship). Lastly, many thanks to Andrew Hendry, Jeff Podos, and Joost Raeymaekers for introducing me to the magical place that happened to be visited, some 185 years ago, by someone named Charles Darwin.