Distribution patterns of invasive Monk parakeets ( Myiopsitta monachus ) in an urban habitat

Distribution patterns of invasive Monk parakeets (Myiopsitta monachus) in an urban habitat.— Several invasive species have been shown to have a marked preference for urban habitats. The study of the variables responsible for the distribution of these species within urban habitats should allow to predict which environmental variables are indicative of preferred habitat, and to design landscape characteristics that make these areas less conducive to these species. The Monk parakeet Myiopsitta monachus is an invasive species in many American and European countries, and cities are one of its most usual habitats in invaded areas. The aim of this paper was to identify the main factors that determine distribution of the Monk parakeet in Barcelona, one of the cities in the world with the highest parakeet density. We defined our model based on eight preselected variables using a generalized linear model (GLZ) and evaluated the strength of support for each model using the AIC–based multi–model inference approach. We used parakeet density as a dependent variable, and an analysis restricted to occupied neighbourhoods provided a model with two key variables to explain the distribution of the species. Monk parakeets were more abundant in neighbourhoods with a high density of trees and a high percentage of people over 65 years. This is interpreted by the fact that parakeets use trees as food sources and support for the nests, and that older people often feed the species. Data support the ‘human–activity’ hypothesis to explain how invasive species can successfully establish in a non–native habitat, and stress how limiting food resources, especially food supplied by humans, may be the easiest way to exert some control on Monk parakeet populations.


Introduction
The spread of exotic species is a major threat to native biodiversity and ecosystem functioning (Diamond, 1989;Temple, 1992;Wilcove et al., 1998;Kolar & Lodge, 2001;Stockwell et al., 2003).The main threats of invasive species can be summarized as: alteration of ecosystem processes (Raizada et al., 2008); decrease of native species abundance and richness through competition, predation, hybridization and indirect effects (Blackburn et al., 2004;Gaertner et al., 2009); changes in community structure (Hejda et al., 2009); and alteration of genetic diversity (Ellstrand & Schierenbeck, 2000).When invasive birds establish in a new habitat, they become part of the local biotic community, and this inevitably has an environmental and economic impact: transmission of avian diseases, or crop damages (Temple, 1992).Identifying the mechanisms that enable these species to establish viable populations in their new areas is one of the most important tools for their future management (Kolar & Lodge, 2001).Elucidating distribution patterns and the key variables that explain these patterns is an advisable approach (Franklin, 2009).
A number of hypotheses have been proposed to explain how invasive species can successfully establish in a non-native habitat.The 'climate-matching' hypothesis assumes that invasive species have a higher probability of successful establishment when they are introduced into regions with a climate similar to that in their native area (Williamson, 1996;Leprieur et al., 2008).In contrast with this, the 'human activity' hypothesis argues that it is human activity that favours this success by disturbing natural landscapes and increasing the number of individuals released and the frequency of introduction (Taylor & Irwin, 2004).These two hypothesis are the best supported by empirical data (Blackburn et al., 2009), andTemple (1992) showed that both mechanisms are conducive to successful reproduction.Some other hypothesis have additionally been proposed, as for instance the 'enemy-release' hypothesis, which states that the abundance or impact of some invasive species is related to the scarcity of natural enemies in the introduced range compared with the native range (Keane & Crawley, 2002;Torchin et al., 2002Torchin et al., , 2003;;Mitchell & Power, 2003).
The Monk parakeet is a common invasive bird species.Due to parrot trade, escapes from captivity, and the accidental or deliberate releases by people that had them as a pet, this species has spread its distribution range from South America (native habitat) to North America and Western Europe (Hyman & Pruett-Jones, 1995;Van Bael & Pruett-Jones, 1996;Cassey et al., 2004).This parakeet is characterized by building many nests within a communal nest structure of tightly inter-twined twigs and sticks (Spreyer & Bucher, 1998;Domènech & Senar, 2006;Burger & Gochfeld, 2009).Each individual nest is called a chamber and is used for breeding and roosting (Domènech & Senar, 2006).Nest structures are often clustered, forming colonies in a group of trees (Burger & Gochfeld, 2009).The nests are used throughout the whole year, and are repaired and reused year after year (Navarro et al., 1992;Eberhard, 1996Eberhard, , 1998)).
Although recent data on Monk parakeets show that number of frost days is a limiting factor for their establishment (Strubbe & Matthysen, 2009), the species has been able to adapt to places with a cold climate, such as Chicago, northern New Jersey, and Brooklyn (Pruett-Jones & Tarvin, 1998).It has also successfully colonised temperate areas, such as in Europe.Human activity therefore also appears to be highly responsible for the distribution of the species (Muñoz & Real, 2006;Strubbe & Matthysen, 2009).Their success as an invasive species provides the opportunity to examine adaptation not only to new habitats and environments, but also specifically to crowded cities (Burger & Gochfeld, 2009).
To date, however, studies on the patterns and key variables responsible for the distribution and successful spread of the Monk parakeet have been based on large geographic areas (Muñoz & Real, 2006;Strubbe & Matthysen, 2009).Although the role of cities is very important in the spread of this species (Pruett-Jones & Tarvin, 1998;Burger & Gochfeld, 2009), available data about the variables responsible for the success of the species within cities is scarce.Barcelona city holds one of the largest populations of Monk parakeets in Western Europe (Domènech, 1997;Domènech et al., 2003).The bird was detected for the first time in 1975 in the Ciutadella Park (Batllori & Nos, 1985) and since then, the species has dramatically increased its populations and has expanded its range throughout the entire city (Clavell et al., 1991;Domènech, 1997;Sol et al., 1997).In 2001, there were 313 nests in Barcelona (Conroy & Senar, 2009), but by 2010 the number had increased to 650 nests, with 1,876 chambers (J.C. Senar & T. Montalvo, pers. obs.).The aim of this paper was to determine the factors that affect the distribution of the Monk parakeet in Barcelona city so as to obtain information about the habitat preferences of the species within the urban habitat.

Study area
The study was conducted in Barcelona city, in northeastern Spain.Barcelona has an area of 102.16 km 2 , 72.34% of which is built up.It is divided into ten districts, which allows decentralized local administration.Each district consists of several 'barrios' (neighbourhoods).Barcelona has 73 neighbourhoods, each with its own personality, historical tradition, and homogeneous urban structure, making them clearly recognizable.Neighbourhoods are hence our sample unit.

Data on species distribution
A census of Monk parakeet was carried out from October 2009 until March 2010.A group of university biology students in the final year of their graduate programme helped us to take a census of the population of Monk parakeets in the whole city.We visited every street and all green areas within each neighbourhood.We marked the exact position of each nest found, obtaining the coordinates for each one.We also recorded the number of chambers each nest contained, and the tree species or the substrate on which it was built.The nest coordinates were later transferred to the GIS system MiraMon, creating a layer of points representing the nests.This layer was combined with a layer of the neighbourhoods, reverting data to the neighbourhood level.The resulting map allowed us to know the number and characteristics of nests within each neighbourhood.
We used the density of chambers (number of chambers per square kilometre) per neighbourhood to estimate population density of Monk parakeets at each sampling unit, following Sol et al. (1997) and Domènech et al. (2003).

Descriptor variables
Demographic, socioeconomic and urban development data were provided by the Statistics Department at the Barcelona City Council (http://www.bcn.cat/estadistica/angles/index.htm).We used variables created by the Department of Basic Information and Cartography to determine the main characteristics of building volumes and spatial distribution.From all these data, we selected those variables that would most likely affect the distribution of the Monk parakeet and that could better describe the characteristics of the neighbourhoods.These variables included: human population density (PD), the percentage of people over 65 years (PD > 65), the percentage of buildings constructed before 1901 (B < 1901), and the percentage of streets and roads (SR).The main reasons for choosing these variables are explained below.Population density (PD) gives us an idea about how many people inhabit in each neighbourhood.This could be a variable of interest since it has been seen to have a strong influence on feral pigeon density (Senar et al., 2009).The percentage of people over 65 years (PD > 65) was considered because this population group seems to be more likely to feed birds in urban habitats (Montalvo & Senar, pers.obs.).The last two variables were related to urban characteristics: the percentage of buildings constructed before 1901 (B < 1901) shows that old buildings were constructed more closely together with each other, leaving insufficient space between them for urban parks and trees; and the percentage of streets and roads (SR) indicates that neighbourhoods with wider streets could have many more green areas and trees where Monk parakeets could nest and feed.Finally, we should point out that as our study was based on a small-scale distribution of the Monk parakeets, it was pointless to choose climatic, topographic or lithologic variables as other studies have done (Muñoz & Real, 2006).
The Department of Parks and Gardens in Barcelona provided additional variables describing the vegetation in each neighbourhood: percentage of grass area (grass area per neighbourhood; GA), percentage of shrub area (shrub area per neighbourhood: SA), density of trees (number of trees per square kilometre; DT), and percentage of forest area (forest area per neighbourhood; FA).Barcelona has 295 species of planted trees, but only a few are used by parakeets for feeding or nesting (Carrillo-Ortiz, 2009).So as to determine whether the species of trees had any effect on Monk parakeet distribution we classified the trees into three groups depending on the type of fruit that they produced: trees with fleshy fruit (e.g.Celtis sp., Ficus sp., Prunus sp., Robinia pseudoacacia), trees with dry fruit (e.g.Populus sp., Tilia sp., Tipuana tipu, Ulmus sp.), and trees with conus or strobilus (e.g.Pinus sp., Cupressus sp., Cedrus sp.).We also included the density of Phoenix sp. and the density of Platanus sp. as independent variables.Phoenix sp. is one of the most commonly used trees by parakeets to build nests in Barcelona (Sol et al., 1997).Platanus sp. is the most abundant tree in the city and is increasingly being used to build nests (Carrillo & Senar, pers.obs.).Densities were measured as number of trees per neighbourhood in km 2 .

Statistical analyses
Because the dependent variables and several of the independent variables did not follow a normal distribution, all the variables were rank transformed (Conover, 1981).We tested the correlation between the different variables to detect any possible colinearity problem.All the correlations were < 0.50, and in most cases < 0.20-0.30.We can therefore be confident that colinearity does not apply to our data.Next, data were introduced in a generalized linear model (GLZ) adjusted to a recent approach based on the Akaike Information Criterion (AIC) (Burnham & Anderson, 2002;Lukacs et al., 2007).The dependent variable (the density of Monk parakeet measured as density of chambers per neighbourhood) and all the explanatory variables were introduced in the model.The eight independent variables were: human population density (PD), percentage of people over 65 years (PD > 65), percentage of buildings built before 1901 (B < 1901), percentage of road system (SR), percentage of grass area (GA), percentage of shrub area (SA), density of trees (DT), and percentage of forest area (FA).Why some neighbourhoods did not have parakeets could be either because the area was not suitable for the species or just that parakeets had not yet colonized it.We thus preferred to restrict our sample size to those neighbourhoods that had already been occupied (i.e.number of chambers > 1).This left us with 51 neighbourhoods (N = 51).
As our survey showed that most nests of Monk parakeets were built in palm trees, we decided to additionally test whether the typology of tree species in an area influenced their abundance.We made a multiple regression approach using the density of chambers as the dependent variable, and the density of Phoenix, the density of Platanus, the density of trees with conus, the density of trees with fleshy fruit, and the density of trees with dry fruit as independent variables.All statistical analyses were carried out using Statistica 10.0 (StatSoft, 2001).

Results
Monk parakeets appeared throughout whole city of Barcelona (fig.1).The total number of detected nests was 650 and the number of chambers was 1876.The 'Sant Pere, Santa Caterina i la Ribera' neighbourhood had the highest concentration of nests (n = 120), most of which were located within the Ciudadela Park.The average density of nests per neighbourhood in 2010 was 8.27 ± 14.69 (n = 73) and that of chambers was 23.51 ± 38.80 (n = 73) (table 1).Variation between neighbourhoods was high.Five neighbourhoods had a total density of chambers of above 80, most had a density of less than 20 chambers, and almost one third (22 neighbourhoods) had no chambers (figs.2, 3A).
The size of the nests, computed as average number of chambers per nest and neighbourhood, was also highly variable.In almost half of the neighbourhoods, nests had an average size of one to four chambers, whereas only seven neighbourhoods had smaller nests (average size < 2 chambers), and nine neighbourhoods had the larger nests (average size > 4 chambers) (fig.3B).
The AIC-based multi-model inference showed that the best model (with the smallest AIC value) regarding the other models consisted of two variables: density of trees (DT) and percentage of people over 65 years (PD > 65) (table 2, fig.4).The next best model included one further variable: density of human population (PD) (table 2).
A multiple regression analysis relating the preference of Monk parakeet for any type of tree did not show any significant result.Despite the abundance of palm trees and Platanus sp. in the city of Barcelona, neither of them affected the density of parakeets (table 3).

Discussion
Information about the environment in areas of origin of invader species may be important to model their distribution in large, invaded geographic areas (Blackburn et al., 2009).However, it has been recently suggested that the non-native distribution of invading species cannot be predicted from the characteristics of the native niche (Petitpierre et al., 2012).Hence, it becomes highly relevant to obtain detailed information about the factors responsible for the distribution of these species in invaded areas.In the case of species that are highly linked to urban habitats, like Monk parakeets (Domènech et al., 2003) or House crows Corvus splendens (Soh et al., 2002) it becomes imperative to predict their preferences and distribution within cities.We found that the model that best explained the abundance of Monk parakeets in Barcelona city showed two relevant variables: density of trees and the percentage of people over 65 years.
The fact that trees are used as a main food source and also to support and build nests (Carrillo-Ortiz, 2009) explains why the density of trees is the main variable to account for Monk parakeet abundance in Barcelona city.Interestingly, our results showed that parakeets had no preferences for different species of trees when selecting preferred 'habitat'.This finding supports a recent study on breeding ring-necked parakeets (Psittacula krameri) in Brussels, Belgium, that showed that parakeets preferred to forage in city parks and gardens due to the huge variety of exotic and ornamental plants (Strubbe & Matthysen, 2011;Clergeau & Vergnes, 2012).Moreover, our finding shows that Monk parakeets in Barcelona currently use other tree species for nesting in cases where palm trees are scarce, which was not the case in the earlier stages of colonization by the species (Sol et al., 1997).In Singapore, Soh et al. (2002) found similar results with another urban invasive bird species, the House crow with regard to the yellow flame trees (Peltophorum pterocarpum).
As the Monk parakeet is a colonial species, the stages of colonization can be a consequence of historical and contagious effects (i.e., the occurrence of one nest may prompt the colonization of that area by other individuals of the population).Nevertheless, the average dispersal distance of the species in Barcelona city is 1,114 ± 190 m (Carrillo-Ortiz, 2009), which provides ample opportunities for the colonization of new areas within the city.The presence of a good number of nests with only one chamber supports that this colonization is taking place.Hence, we think that our data reflect habitat selection in spite of historical and contagious effects.
Our data support the 'human-activity' hypothesis to explain how invasive species can successfully establish in a non-native habitat.We found that the percentage of people over 65 years old was an important variable in the best fit model, and the density of human population was strong in the second model (AIC < 2).Human presence thus clearly favours the invasion by Monk parakeets, not only because human activity favours escapes and releases of individuals, but also because of the foraging advantage that individuals enjoy because of (older) people supplying food.Monk parakeets have changed their foraging behaviour in recent years, adapting to novel conditions and using novel resources of food offered by humans (Lefebvre et al., 2004;Sol, 2007).A recent report showed that these new food resources in Barcelona are mainly cereals and bread, and they account for up to 40% of parakeet food sources (Carrillo-Ortiz, 2009).The population group most likely to feed birds is usually retired people (Montalvo & Senar, pers. obs.).Similarly, Pithon (1998) observed that parakeet densities in the UK correlated highly with densities of detached and semi-detached houses inhabited by retired people.
Taken together, our results support the view that the preference and success of invasive species in cities is related to the abundance of food available and nesting sites in these anthropogenic habitats, so that invasive species can easily increase both breeding success and parental survival (Chamberlain et al., 2009).As food sources play a major role in favouring invasive species in general and Monk parakeets in particular, we propose that limiting food resources provided by the population would be a reasonable way to control these species in urban habitats.However, further studies with new variables are needed to understand better and determine why Monk parakeets prefer these areas in Barcelona city.

Fig. 1 .
Fig. 1.Distribution map of Monk parakeet nests in the neighbourhoods of Barcelona city.The dots depict the exact localization of the nests.

Fig. 2 .
Fig. 2. Distribution map of Myiopsitta monachus in Barcelona city.The presence of Monk parakeet measured as the density of chambers (number of chambers per square kilometre) in the different neighbourhoods, where the species was recorded, is categorized in five classes according to a grey scale.

Table 1 .
Descriptive statistics on the density of nests and the density of chambers of the Monk parakeet in Barcelona, both in all neighbourhoods (n = 73) and restricted to neighbourhoods with at least one chamber (n = 51).Density is measured as number of nests or chambers per square kilometre.

Table 2 .
Results of the multimodel inference using AIC to predict which variables in Barcelona city showed the best approximating model that explained the density of Monk parakeets (measured as number of nest chambers): PD.Human population density; PD > 65.Percentage of people over 65 years; B < 1901.Percentage of buildings constructed before 1901; SR.Percentage of streets and roads; GA.Percentage of grass area; SA.Percentage of shrub area; DT.Density of trees; FA.Percentage of forest area.