Urbanization is associated to a loss of phylogenetic diversity of birds in a medium size city on the Andes of Colombia, South America

ABSTRACT Accelerated rates of urbanization negatively impact natural habitats and biodiversity. Studies of the effects of urbanization on fauna have emphasized taxonomic diversity, but this does not necessarily reflect effects on the evolutionary history of assemblages. Here, we study how urbanization influences phylogenetic diversity (PD) of bird assemblages in a small–medium city in the Central Andes of Colombia. Between 2016 and 2017, we performed bird surveys in 76 urban cells and 23 rural cells around the city. Then, we measured habitat characteristics (e.g. anthropogenic noise level, impervious and forested area) in sampled cells. PD of birds was between 1.37–1.42 MY higher in the rural than in the urban habitat. Within the urban habitat, the PD of birds reduces as impervious area and noise levels increase, while in the urban cells with most forested areas reach the highest PD values. Most of these differences and relationships disappear when we performed similar analysis controlling for differences in species richness. A low turnover of PD between assemblages of birds recorded inside Armenia and its less urbanized surroundings suggest green areas in the study area could mitigate the effects of urbanization in native fauna, and ameliorate connectivity between urban cells.


Introduction
Urban areas are spreading at a fast pace and are considered one of the most significant human induced changes in natural habitats, leading to a substantial loss of biodiversity (Aronson et al. 2014;Forman 2014).Among the many habitat changes associated with urbanization, the reduction of forested areas and the increase in impervious areas dramatically alter the availability of vital resources such as food and nesting places for many species (McKinney 2002;Shanahan et al. 2014).Moreover, wildlife in cities is frequently subjected to an increase in stress levels, mortality by novel predators, and the collision with buildings and vehicles (Sims et al. 2008;Kight & Swaddle 2011;Loss et al. 2013;Santiago-Alarcon & Delgado-V 2017).In other words, environmental characteristics in urbanized places impose filters that only species with specific traits can overcome in order to establish in those novel habitats (McKinney 2006;Croci et al. 2008;Sol et al. 2014;Aronson et al. 2016).
Research about how urbanization impacts faunal communities has primarily been focused on assessing the decrease in taxonomic diversity (e.g.McKinney 2008;Luck & Smallbone 2010;Magura et al. 2010;Sol et al. 2017a;Orlando et al. 2020;Bhakti et al. 2020;Zuñiga-Palacios et al. 2020).However, approaches based on taxonomic diversity consider all species as equal and fail to recognize differences among them.Therefore, taxonomic diversity alone does not necessarily reflect all the potntial harmful effects of urbanization processes on the evolutionary history of lineages that compose faunal communities, or even on the ecological role that species play in a given ecosystem (Lean & Maclaurin 2016;Morelli et al. 2016;Palacio et al. 2018a;Freitas & Mantovani 2018).At present there is a plethora of metrics to quantify the link between differences among species and diversity (Vellend et al. 2011;Swenson 2014), but the one suggested by Faith (1992) is the more frequently used and best evaluated (Chao et al. 2010;Scherton & Faith 2018).This author proposed that the phylogenetic diversity (PD) of an assemblage is the sum of all branch lengths in the portion of a phylogenetic tree connecting the focal set of species.Since the phylogenetic distance (given by branch lengths in a tree) between species can be used as a proxy for the magnitude of feature divergence, and hence niche similarities and intensity of competition between them (Darwin 1859;Tucker et al. 2018), PD offers insights for understanding factors shaping community structure and species coexistence (Webb et al. 2002;Cadotte et al. 2010).
PD has offered important insights into community ecology, and also into conservation biology (Pellens & Grandcolas 2016;Scherton & Faith 2018;Véron et al. 2019).Since more species in an assemblage add more branches to the phylogenetic tree connecting them, there is a positive relationship between species richness and PD (Kluge & Kessler 2011;Swenson 2014;Tucker et al. 2018).Therefore, when urbanization reduces the taxonomic diversity of assemblages, this can trigger a reduction in PD, that is, a loss in the amount and divergence of the evolutionary history of species within the assemblages (Ibáñez-Álamo et al. 2017;Sol et al. 2017a;Palacio et al. 2018a;La Sorte et al. 2018).The inclusion of evolutionary history in studies of urbanization, and conservation biology as a whole, offers information about whether species have the capacity to evolve adaptations in response to human induced potential threats and therefore, provide baseline data to improve decision-making both in natural and in urbanized areas (Lankau et al. 2011;Winter et al. 2013;Edwards et al. 2015).Unfortunately, most of the studies addressing the relationship between urbanization and PD have been performed in temperate regions (Lancaster & Rees 1979;Beissinger & Osborne 1982;Jokimäki & Kaisanlahti-Jokimäki 2003;Clergeau et al. 2006;Garden et al. 2006;Luck & Smallbone 2011;Čeplová et al. 2015;Morelli et al. 2016), in contrast to the Neotropics, where very few studies have applied the aforementioned approach (Lees & Moura 2017;Ibáñez-Álamo et al. 2017;Sol et al. 2017a;Silva-Junior et al. 2018).Given that urbanization-wildlife dynamics seem to vary between different cities and regions of the world, it is necessary to carry out more studies in regions that represent reservoirs of biodiversity, such as in the Neotropics (Luck & Smallbone 2010;MacGregor-Fors & Escobar-Ibáñez 2017).
The Andes of Colombia, South America, is one of the biggest hotspots of biodiversity (Myers et al. 2000;Pauchard & Barbosa 2013), and many species have limited geographical ranges of distribution, with speciation processes occurring at particularly high rates (Kattan et al. 2004;Cadena et al. 2011;Hutter et al. 2017).In this region, however, there is a high unplanned rate of habitat transformation, including urbanization, that is threatening biodiversity (Andrade-Medina & Bermúdez Cárdenas 2010;Quintero-Gallego et al. 2018).In a previous study, Carvajal-Castro et al. (2019) compared bird species diversity between urban habitats (inside the city) and rural habitats (the less urbanized surroundings) in Armenia, a city located in the coffee region of the Central Andes of Colombia, characterized by several biological corridors composed by remnants of secondary forests and Neotropical giant bamboo patches (Guadua angustifolia, guadual).These authors found an overall pattern of highest species diversity of birds in the rural habitat; moreover, inside the urban habitat, cells (places of 250 × 250 m) with higher abiotic noise intensity and higher impervious area exhibit less diversity than cells with higher guadual and forested areas.In concordance with other studies performed in the same locality (Vanegas-Guerrero et al. 2016), data with birds suggested that forested patches could play an important role in the conservation of native tetrapods.However, Carvajal-Castro et al. (2019), did not consider the phylogenetic relationships and evolutionary patterns of bird species in such assemblages.
Here, we study how urbanization could affect the PD of bird assemblages in the city of Armenia, Colombia.Addressing these effects is important to obtain baseline data that allows a better understanding of the effects of urbanization on biodiversity in the Neotropical realm.
To do this, we first compared the PD between urban and rural habitats and calculated the phylogenetic turnover of bird assemblages between both habitats.Given the well-known positive relationship between species richness and PD (Kluge & Kessler 2011;Swenson 2014;Tucker et al. 2018), and a reduction in taxonomic diversity from rural habitats toward the city in the study area (Carvajal-Castro et al. 2019), we expected that urban habitats exhibit a lower PD than rural habitats.Second, we tested the relationship between habitat characteristics (e.g.anthropogenic noise intensity, and percentage of impervious and forested areas) and the PD of birds among 76 places (hereafter urban cells) within the city of Armenia, and 23 places (hereafter rural cells) outside the limits of the city.We predict that urban and rural cells with higher forest cover and lower impervious areas or lower open (deforested) areas will exhibit higher PD than places where forest is scarce and impervious or open areas predominate.Similar intra-city scale analyses of the effects of urbanization on PD of birds are scarce (Morelli et al. 2016;Ibáñez-Álamo et al. 2017;Sol et al. 2017a).

Study area
The city of Armenia is a medium-sized city located in the Central Andes, Colombia, in the northern Andes of South America (4.51°N, 75.66°W; 1483 m elevation).This city encompasses places with a predominance of impervious surfaces (i.e.places with high levels of urbanization), and others with a predominance of forest remnants and bamboo patches (hereafter guaduals) (Figure S1A, B in the online supplemental material).This city is located in a landscape with isolated remnant forests embedded in a matrix of grasslands used for human activities as livestock and agriculture (Figure S1C, D).An aerial image and a detailed history of habitat transformation in Armenia are presented in Carvajal-Castro et al. (2019) and Ljm et al. (2019).

Bird surveys
The city of Armenia was divided in 364 urban cells of 250 × 250 m each.We made bird surveys in urban cells separated by 250 m to reduce chances of pseudoreplication (i.e.counting an individual more than once).Because of this, we sampled only 76 urban cells for bird diversity.We made bird surveys only on sunny days between 27 September 2016 and 17 February 2017, and using standardized techniques according to Gregory et al. (2004).Bird surveys were conducted by two observers using binoculars for 20 min.We sampled 1-4 urban cells per day between 0600 h and 0800 h; we sampled each urban cell only once during the whole study.We established two orthogonal transects of 100 m each from the center of each urban cell; one of these two transects was randomly selected and the avian surveys performed in it.We recorded species and number of individuals per species present inside a distance ˂ 50 m from the central line of a transect selected in each urban cell.For the rural habitat, we made bird surveys in 23 rural cells (250 × 250 m each one) between 29 November 2016 and 17 February 2017, using the same methods described for urban cells.We based our bird counts only on visually observed individuals.Also, because we were unable to perform surveys inside forest remnants within Armenia, our samplings were performed from areas outside forested patches.All bird species were identified using Hilty et al. (1986) and McMullan et al. (2011).
There was a mismatch between our survey period (i.e.September-February) and the period where migratory birds arrive (i.e.October-March) in the study area (Gill 1995;RESNATUR 2004).Therefore, the absence of migratory birds in some sampling cells could be because they have not arrived yet at the time of our surveys, and not because of potential effects of habitat characteristics.To control for possible biases generated by this temporal mismatch we did not include migratory species in this study.However, since migratory species would increase PD by increasing the number of species, and by adding new functions and evolutionary history to the sites they arrive, we provide a summary of the number of species migrants in each habitat type (Table S1).

Habitat characteristics
Values of habitat characteristics used in this study are those we obtained in a previous study (Carvajal-Castro et al. 2019).We measured five habitat characteristics for each urban cell with bird surveys using QGIS 2.16 (QGIS Development Team 2016): straight distance to the nearest boundary of the city, the percentage of impervious area, percentage of guadual area, percentage of forested area, and percentage of open area (e.g.grasslands or soccer fields).Moreover, anthropogenic noise intensity (hereafter noise) was measured using a sound level meter Extech 407,730 configured in C function and located at 1.5 m height.Noise intensity for each urban cell was the mean value obtained from four measurements made at the extremes of the two orthogonal transects established in each sampled cell.These noise measurements were performed during bird surveys (i.e. between 0600 h-0800 h).Noise was measured in decibels (dB), but statistical analyses were performed after converting such dB values to a linear scale (pressure, Pa).Habitat characteristics were measured in 23 rural cells with the same software tools mentioned for urban cells; however, in rural cells the percentage of cultivated areas (mostly coffee) was included as a habitat characteristic.We did not measure the level of anthropogenic noise in the rural habitat, but it was clearly lower than in the urban habitat.

Phylogenetic relationship of birds
Our analyses were based on the phylogeny for birds published by Jetz et al. (2012), which is the most up to date phylogeny and has the greatest taxonomic coverage.To take into account phylogenetic uncertainties, a mean of PD indexes was calculated using 1000 timecalibrated Bayesian posterior phylogenies randomly sampled from www.birdtree.org(Hackett backbone; Jetz et al. 2012).Since this phylogeny places species without genetic information using taxonomic information, which could increase uncertainty in PD calculation (Rabosky 2015), for this analysis we only considered species with genetic information.We left out of the analyses species that were not present in phylogenies (8/100 species in total).

Data analysis
For every cell we calculated the PD using the indexes based on Hill numbers at q = 0 and q = 1 (hereafter PD0 and PD1, respectively) proposed by Chao et al. (2010).These indexes include different levels of sensitivity to phylogenetic distances between species and its relative abundance.When q = 0, the calculation of PD does not take in account differences in the relative abundance of species; when q tends to 1, species are weighted according to their relative abundance.All of the mathematical calculations needed for this PD index were performed using the estPD function of iNextPD package (Hsieh & Chao 2017) in the software R (R Core Team 2018).We use Hill numbers for testing PD because it accomplishes with the replication principle, which states that 'N equally large, equally diverse groups with no species in common, the diversity of the pooled groups must be N times the diversity of a single group' (Jost 2006(Jost , 2007;;Chao et al. 2010).Many authors have pointed that diversity metrics that do not satisfy the replication principle give erroneous and illogical results (see references in Chao et al. 2010).There are many other PD indices with different properties; however, many such metrics are highly correlated, in some cases are mathematically identical, and do not satisfy the replication principle (Chao et al. 2010;Swenson 2014).For a deeper discussion of index properties based on Hill numbers see Jost (2006Jost ( , 2007)), Chao et al. (2010Chao et al. ( , 2016)), and Chiu et al. (2014).
PD can be highly correlated with species richness (Swenson 2014).Hence, variation in values obtained from PD indexes can indicate a change in evolutionary history of assemblages because of a change in species richness or a change of evolutionary history alone (i.e.regardless of species richness).Therefore, after corroborating such positive relationship for our data with a Pearson correlations test (Figure 1), we calculated a standardized effect size (SES) (Webb et al. 2002) for our PD indexes (hereafter sesPD0, sesPD1), using a 1000 times randomization of the community matrix under the independent swap algorithm (Gotelli 2000) in our null model with Picante package (Kembel et al. 2010) in R.This algorithm randomizes the community matrix conserving the columns and rows total sums across all randomizations, and it applies constrained randomization preventing error type I.In addition, it is used for the presence-absence matrix but does not randomize trait data (Gotelli & Entsminger 2001;Swenson 2014).
We compared the PD (sesPD) between urban and rural habitats using Student's t-tests.Additionally, to control for possible effects of differences in the number of sampled cells, we repeat the analysis randomly selecting 23 urban cells to compare with the 23 rural cells.We did this comparison 1000 times to ensure the comparison robustness.We also calculate the index of evolutionary distinctiveness (ED).This index quantifies the evolutionary uniqueness of species composing a given assemblage or the relative contribution of one species to PD (Isaac et al. 2007).For this, we used subtrees only containing the species in each urban and rural sampled cell.This calculation was performed in Picante packages in R. We compared ED values between rural and urban habitats through a Wilcoxon text.Finally, we calculated the beta-PD to test the phylogenetic turnover of bird assemblages between both habitats.This beta metric was calculated under a Hill number framework using q = 0 (unweighted by abundance), and q = 1 (weighted by abundance) (Chiu et al. 2014;Chao et al. 2019).Beta-phylogenetic metric varies between 1 (indicates the lowest beta-PD; i.e. that compared assemblages are similar) and N, with N being the number of assemblages compared in the analysis (if beta-PD equals N, this indicates that the compared assemblages are totally dissimilar); in other words, beta-PD in this study can vary between 1 and 2. All these calculations were performed with h hillR package in R (Li 2018).
In order to test relationships between habitat characteristics with PD, we performed the following statistical analyses.First, redundant variables representing habitat characteristics were reduced by conducting a principal component analysis (PCA) applying a Varimax-rotation in SPSS v. 21 (IBM Corp N 2013).Variation in the measured variables was summarized in three principal components (PCs; see Table S2 and S3 in supplementary material).Overall, the three PCs were mainly correlated in the urban habitat as follows: PC1, impervious and guadual area; PC2, distance to the boundaries of the city, and the amount of open area; and PC3, forest area.In the rural habitat, the three PCs were PC1: amount of open and plantation area, PC2: forest and impervious area, and PC3: distance to the boundaries of the city and guadual area.Then, we used the resulting PCs as new variables of habitat characteristics and related them to phylogenetic diversity values using a general linear model (GLM) with a Gaussian error distribution.For all of these analyses, the sampling unit was the sampled cell (urban or rural).

Phylogenetic diversity patterns in the urban habitat
The GLMs (Table 1) showed, first, that the PD (at order q = 0 and q = 1) of birds decreased as the impervious surface area increases.Second, there was not a significant relationship between distance to the boundaries of the city and the amount of open area with PD.Third, as the amount of forest area increases, the PD (at order q = 0, not at order q = 1) increases.Fourth, noisier urban cells have lower bird PD than quieter urban cells (at order q = 0, not at order q = 1).When differences in the number of species among urban cells are considered using the standardized PD (sesPD) for testing these results, only the inverse relationship between amount of impervious area and PD at order q = 1, remains significant (Table 1).

Phylogenetic diversity pattern in the rural habitat
In the rural habitat, bird PD at any order (i.e.q = 0, q = 1) was unrelated with the habitat characteristics measured as principal components (i.e.PC1, PC2 and PC3; Table 2).A similar result was obtained for the sesPD, except for the diversity at order q = 0 that increased with the amount of open and plantation area (Table 2).Similar to the urban habitat, the amount of forest area tends to relate positively with PD (at order q = 1), but this tendency was not statistically significant (Table 2).

Discussion
Urbanization can reduce PD through the loss in species number or due to a reduction in species evolutionary history (Sol et al. 2017a).Our results indicate that the lower PD of birds in the urban habitat (city of Armenia) than in the rural habitat (less urbanized surroundings of Armenia) is mediated by a decrease in species richness rather than by evolutionary history alone.The reduction of species richness with urbanization in the study area also may explain, at least in part, the lower ED in the urban habitat than in the rural habitat (Sol et al. 2017a, b).At the intra-city level analysis, our results also show a reduction in evolutionary history as the impervious area increases regardless of differences in species richness.The effect of noise (negative) and forest area (positive) on bird diversity become absent when the relative abundance of individuals and differences in species richness are taken into account.
The reduction of PD as the urbanization level increases in the study area is attributed to a process of habitat filtering (McKinney 2006;Sol et al. 2014;Morelli et al. 2016).Since PD and functional diversity (based on seven functional traits; unpublished data) were positively correlated for our study system (r = 0.90, P < 0.001 at order q = 0; r = 0.74, P < 0.001 at order q = 1), environmental changes associated with high levels of urbanization, excluding taxa whose phenotype shifts from an optimal trend, can reflect a reduction of PD.For instance, reviews of the urban effects on native fauna (Chace & Walsh 2006;Escobar-Ibáñez & MacGregor-Fors 2017;La Sorte et al. 2018) have determined that specific functional trophic groups of birds are more prone to survive in urbanized habitats of Latin American: omnivores, granivores, and insectivores.Further analyses of functional traits of birds in the study area are necessary for a detailed understanding of the processes and mechanisms underlying the effects of urbanization in native species, but results in our study agree with these tendencies.The most abundant species in the city of Armenia were insectivores and granivores: Columbina talpacoti, Myiozetetes cayanensis, Pitangus sulphuratus, Pygochelidon cyanoleuca, Pyrocephalus rubinus, Sicalis flaveola, Troglodytes aedon, Tyrannus melancholicus, and Zenaida auriculata (Hilty & Table 1.Summary of the results obtained from the general linear models between phylogenetic diversity (at different sensibility to relative abundance of individuals: q = 0, q = 1) and characteristics in urban cells in the city of Armenia, Colombia.Table 2. Summary of the results obtained from general linear models between phylogenetic diversity (at different sensibility to relative abundance of individuals: q = 0, q = 1) and characteristics in rural cells around Armenia, Colombia.Brown 1986; Del Hoyo et al. 2019).The habitat filtering imposed by urbanization can also be related to other ecological characteristics of birds such as nesting site (MacGregor-Fors & Escobar-Ibáñez 2017).
The low turnover of PD between assemblages of birds recorded inside Armenia and its less urbanized surroundings suggest that some elements in the city may be buffering the effects of urbanization in the biodiversity of birds.Armenia has numerous forest remnants and bamboo patches that cover ~ 30% of the city area (CRQ-SIG 2016), which could enable birds to move between rural and urban habitats, reducing the level of phylogenetic turnover.It is unknown whether forested corridors and other green areas crossing Armenia are facilitating the movements of bird species through the urban matrix, and hence promoting population connectivity between urban and rural habitats; therefore, this hypothesis needs to be tested (Leveau & Zuria 2017).However, several studies have found that intermediate transformed environments could act as a buffer against loss of PD (Frishkoff et al. 2014;Sol et al. 2017a).Moreover, the habitat around Armenia is highly disturbed, and taxonomic and PD analysis (Carvajal-Castro et al. 2019; this study) reveals that birds use forested areas, open and cultivated areas.In other words, species in the rural habitats can deal with some level of disturbance in the habitat and possibly can move to some places inside the city, which also reduces our beta diversity values.
Regarding patterns inside the urban habitat, PD decreased as the impervious area increased.Similar results have been found in other cities both in tropical and temperate regions (Morelli et al. 2016;Ibáñez-Álamo et al. 2017;Sol et al. 2017b).Of the several habitat characteristics we measured in urban cells, the amount of impervious area seems to be the most important factor affecting negatively the biodiversity of birds.In fact, it was the only one whose effect remains statistically significant after considering the relative abundance of species and when the difference in species richness is considered (i.e.sesPD).This implies that higher levels of urbanization in the study area produces a reduction in PD due to both the loss of species number and the reduction in species evolutionary history (Sol et al. 2017a).Possibly, this occurs because an increase in impervious area (i.e.intense development of cities) is associated with diverse nonexcluding factors reducing fitness of birds; for instance, a reduction of nesting sites and food availability, and the increase in exposure to novel diseases, stress levels, predation risk, and mortality because of collisions against windows of buildings (Partecke et al. 2006;Delgado-V & French 2012;Wittig et al. 2017;Santiago-Alarcon & Delgado-V 2017).
As forest area coverage increases, the taxonomic diversity of birds increases in the city of Armenia (Carvajal-Castro et al. 2019).Something similar happens with PD in the urban habitat, when calculated at q = 0 (i.e.unweighted by abundance).However, this statistically relationship disappears when PD is calculated at q = 1 (weighted by abundance).Since our surveys were performed in open and impervious areas, it is expected that we overlooked or recorded few individuals of birds dependent on forested habitats (e.g.Accipiter striatus, Eupsittula pertinax, and Myioborus miniatus).Therefore, it is possible that forest remnants contribute to the PD of bird assemblages in urban cells because they favor the presence of species with low abundance, but when forested areas are absent, the loss of such lowabundant species does not impact too much on diversity as they contribute slightly to the index computation.When differences in species richness are included in metrics of PD (i.e.sesPD), the relationship between PD and forest area disappears at any order q.
Noise has been related negatively to taxonomic diversity and PD0 in Armenia (Carvajal-Castro et al. 2019;this study), and other cities in tropical and temperate regions (Fontana et al. 2011;Proppe et al. 2013;Lepczyk et al. 2017).It is possible that high levels of noise pollution reduce the number of species living in a given place because it masks the acoustic communication signals, and may cause a reduction in immune responses associated with high levels of stress (Kight & Swaddle 2011;Halfwerk et al. 2018, but see Moiron et al. 2015;Sánchez-González et al. 2020).In contrast, the species that are not filtered by noise may be overrepresented by individuals as a result of a lower interspecific competition; therefore, the absence of a relationship between noise and PD at q = 1 can be explained by a partial compensation of PD loss through an increase in species evenness (Francis et al. 2009;Sol et al. 2017a).These acoustic filtered species are not necessarily clustered within a phylogeny.In a study with bird assemblages in a small to medium sized city in Mexico (area = 64 km 2 ; human population = 500,000 inhabitants), birds forming dawn choruses in highly noisy places exhibit lower species richness (17 species) compared with those in significantly quieter places (59 species) (Marín-Gómez et al. 2020).These authors, however, pointed out that bird dawn choruses were composed by phylogenetically unrelated species and there was no phylogenetic clustering in noisy places as expected.
It is feasible that rural habitats play a role as a source habitat for some of the species recorded within cities.Hence, with increasing distance from that source within cities (and with increased intensity of urbanization), more species may be filtered out with only some remaining in the inner areas of the cities.According to this, an inverse relationship between distance to a habitat source and species richness has frequently been found in studies of biogeography, community ecology and conservation biology (Harris & Harris 1984;MacArthur & Wilson 2001;Krasnov et al. 2005;Vanschoenwinkel et al. 2007).However, our data for birds in Armenia did not show a negative relationship between species diversity and distance to the boundaries of the city.This may be because most places around Armenia are open areas and plantations; hence, bird assemblages in them tolerate disturbances, which could favor movement of species between the rural and urban habitat.As mentioned above, this hypothesis needs to be tested, but it predicts that birds in the rural habitat tend to exhibit ecological traits similar to those in species that survive in cities.For instance, like in the urban habitat of Armenia, and other cities in Latin America (Chace & Walsh 2006;MacGregor-Fors & Escobar-Ibáñez 2017;La Sorte et al. 2018), among the most abundant species recorded in the rural habitat were insectivores, granivores, and omnivores: Pionus menstruus, Pygochelidon cyanoleuca, Sicalis flaveola, Thraupis episcopus, Turdus ignobilis, and Zenaida auriculata (Hilty & Brown 1986;HWB 2019).
We are aware of the limitations in the present study.First, inside forested patches of the study area there are bird species that we did not record in this study (Marin 2005), likely because our surveys were performed from open areas.This bias could explain why we recorded a lower species richness in Armenia in comparison to that recorded for other cities in Latin American (Ortega-Álvarez & MacGregor-Fors 2011; Escobar-Ibáñez & MacGregor-Fors 2017).However, this systematic bias does not invalidate our results.This is because our sampling of bird diversity would be more exhaustive in highly urbanized cells with lower forest coverage than in urban cells with lower urbanization intensity and more forest coverage.The species in the forests we have not recorded turns such forest remnants into places of concern for conservation.Second, our sampling was limited to visual records, which can underestimate species richness.However, the use of mist-nest or other techniques complementing records based on visual surveys was not feasible under the logistic conditions of urban cells.Moreover, the detectability of birds in a ratio of 50 m seems to be good enough for spatial comparisons of community structure in urban habitats and open areas like those present in the rural habitat (van Heezik & Seddon 2012, 2017).In fact, our sampling coverage for both habitats was > 0.90 (Carvajal-Castro et al. 2019).Finally, replication is necessary; that is, similar studies in other cities of the Andes should be performed to establish if patterns recorded in this study are widespread.Moreover, considering the limited temporal scale of this study, further sampling is necessary to cover aspects related to some birds' phenology and activity patterns (e.g.climatic seasonality and variation of resources offered by plants), and the expected accentuated loss in bird diversity as the urbanization in the study are continue to growth.
The effects of urbanization on biodiversity are not exclusive to birds; such effects may be even stronger for other organisms with less vagility and more sensitive natural history traits that constrain their presence in gray areas of urban habitats (e.g.plants, mollusks, amphibians, reptiles) (Elmqvist et al. 2013;McCleery et al. 2014).However, since birds play an important ecological role in ecosystems, and they are good 'umbrella species' for other taxa (Rodrigues & Brooks 2007;Sekercioglu et al. 2016), management strategies benefiting bird diversity in cities are expected to favor other species in urban habitats.Green areas in Armenia likely offer diverse resources to be exploited by species and could ameliorate the effects of urbanization on native fauna (MacGregor-Fors et al. 2016;Faggi & Caula 2017;Zuñiga-Palacios et al. 2020).Therefore, actions like promoting the connectivity of the green areas and the presence of native plant species could buffer the loss of PD of birds and other organisms as cities continue to grow (Threrlfall et al. 2017).These actions would benefit not only resident bird species, but also migratory species that use urban habitats in cities of Colombia (Muñoz et al. 2007;Ayerbe-Quiñones et al. 2009;La Sorte et al. 2014;Zuckerberg et al. 2016;Rosselli et al. 2017;Palacio et al. 2018b).

Figure 1 .
Figure 1.Relationship between taxonomic diversity and phylogenetic diversity for the urban habitat (upper plots, n = 76 cells) and the rural habitat (bottom plots, n = 23 cells).Diversity values calculated at different sensibility to relative abundance of individuals (q = 0, q = 1); see text for details.

Figure 2 .
Figure 2. Comparisons of phylogenetic diversity (PD) before and after considering differences in species richness (sesPD) between the urban (n = 76 cells) and rural habitat (n = 23 cells).Phylogenetic diversity is calculated based on Hill numbers and at different levels of sensitivity to species relative abundances (PD = 0, PD = 1).See text for details.The illustration shown an individual of Momotus aequatorialis, a charismatic bird present both in the urban and rural habitat; this species is frequently used in the study area as a symbol of conservation, environmental education plans, ecotourism, and business (illustration by María José Tovar-Gil).