Study area

Mexico City is the capital of Mexico and is located at 19.4326° N and 99.1332° W. It is part of the Mexico City Metropolitan Area (MCMA), which includes Mexico City plus two states (State of Mexico and Hidalgo). Here, only Mexico City is included to generate information on the urban forest characteristics that can be incorporated into land use planning and administrative instruments that only operate in Mexico City, as the other states of the MCMA have independent governments. Mexico City has 9.21 million inhabitants (^{National Institute of Geography and Statistics [INEGI], 2020}). It has a surface of 1493 km², of which, 42% corresponds to developed land and 58% is conservation and agricultural land (Fig. 1a). The city is located at 2240 meters of elevation, and it falls under the biogeographical province termed the Trans-Mexican Volcanic Belt (^{Morrone, 2010}); here, native tree species are species native to this province. The climate in the urban area is temperate with dry winter climate (Cwb), with an average temperature of 16.6 °C (maximum average of 27 °C from March to May, minimum average of 3 °C from November to January), and an average annual precipitation of 625 mm (rainy season from June to September).

The location of Mexico City within Mexico is shown in the upperleft inset.

Figure 1 Map of the study area showing the distribution of a) sampling plots distributed across the urbanized area of Mexico City, b) land use types, and c) boroughs.  

Mexico City’s land-use types and boroughs

The landscape of Mexico City comprises two major land uses, the “urban area” (42%) and the “conservation ground” (58%) (Fig. 1a). In this research, the focus was on the urbanized area and the conservation ground was not included. The “urban area” is defined as areas influenced by a high human population density and activity, a large proportion of build-up and impervious surfaces, vehicular traffic, and a variety of industrial activities (^{Ministry of Urban Development and Housing [Seduvi], 2003}), and was delimited here by census tract polygons of INEGI. The “conservation ground” is defined as protected areas intended to maintain the biodiversity of the natural ecosystems and preserve their ecosystem services (^{Rivera-Hernández, 2016}); is distributed in the south part of the city and includes natural vegetation, rural areas, and agricultural lands. Even though there are natural protected areas within the “urban area” land use, such as Cerro de la Estrella, Ejidos de Xochimilco and San Gregorio, among others, the category of “conservation ground” distinguishes the urbanized from the non-urbanized area of Mexico City.

Within the urban area, land use classification includes 26 types (with a total of 141 sub-categories) (^{Seduvi, 2003}). To simplify land use information and have categories than can be comparative with other cities, all land use types were summarized into six types (Fig. 1b) as follows: residential (33.7% of the urban area), mobility networks (18.6%), mixed residential-commercial (15.3%), green areas (15.0%), services/urban equipment (8.9%), industry (2.7%), and the remaining 5.8% is regulated through Partial Programs of Urban Development (i.e., indigenous territories) (^{Seduvi, 2003}). Land-use information was obtained from the Ministry of Urban Development and Housing of Mexico City (Seduvi). The 16 boroughs of Mexico City were analyzed and the cut of the urbanized area was applied in the spatial polygon leaving outside of the analyses the conservation ground.

Field methods

A total of 500 sampling plots (Fig. 1a) were placed using stratified random sampling throughout the urbanized area of Mexico City. Studies of urban forest structure and diversity frequently used an average 200 plot locations (^{Nowak & Crane, 2000}; ^{Escobedo et al., 2010}), thus, a sampling size of 500 plots for a large city such as Mexico City was considered sufficient and it has been reported in other studies (^{Toledo-Garibaldi et al., 2023}, ²⁰²⁴). Stratification was done utilizing several layers representing land use, land cover, canopy cover, and population data, and was used to inform sampling design and reduce bias in the data. The number of sampling plots per land use type ranged between 8 and 189, and per borough ranged between 9 and 107 due to the different surface area covered by each land use type and borough (Supplementary material 1), with smaller areas having fewer sampling plots and larger areas having more sampling plots. For instance, within the study area, the boroughs smaller than 3000 ha had up to 10 plots, while boroughs larger than 6000 ha had more than 40 plots.

A circular plot size of 400 m² was selected because is commonly utilized in urban forest assessments (i.e., United States Forest Service, ^{Nowak & Crane, 2000}). Within each fixed area plot, all trees and shrubs with a diameter at breast height (DBH) ≥ 5 cm were identified to the species level and their DBH, tree height and canopy width were measured (^{Nowak & Crane, 2000}). Canopy width was measured with a metrical tape, averaging two perpendicular measurements of the tree crown. Data from field samplings was used to derive urban forest compositional and structural variables.

Urban forest composition and structure

In total 12 urban forest compositional and structural variables were derived and aggregated at the plot level: (1) species richness, (2) native species, (3) introduced species, (4) species of temperate-affinity, (5) of tropical affinity, (6) of sub-tropical affinity, (7) evergreen species, (8) deciduous species, (9) basal area (BA; m²/ha), (10) canopy cover (m²), (11) number of trees, and (12) tree height (m).

Species richness is a common trait used to describe vegetation communities and urban forests, and it allows making comparisons between cities (^{Staudhammer et al., 2018}). Native species were those naturally distributed within the biogeographical province of the Trans-Mexican Volcanic Belt (^{Morrone, 2010}), knowing the proportion of native vs non-native species shades light on the ecological integrity of the urban forest in terms of its functions and the associated species (Ordóñez & Duinker, 2013). Evergreen species are those that retain their leaves all through the year, including both gymnosperms and angiosperms, while deciduous species are those that shed most of their leaves in a specific season (^{Rzedowski, 2006}). Biogeographical origin (tropical, sub-tropical, temperate) was assigned to each species based on the available floristic databases (i.e., United States Department of Agriculture, Missouri Botanical Garden, Conabio) and floristic inventories of the region from ^{Rzedowski (2005)}. The biogeographical origin provides information about the climate and original habitat of certain species (^{Pataki et al., 2013}), and can be related to species' performance, ecosystem services, and the preferences of urban residents (^{Pataki et al., 2013}).

To further characterize the urban forest, species abundances, and DBH classes were described for all sampled trees. The important value index (IVI) was calculated to determine the species’ dominance. The index considers abundance (number of individuals per species), dominance (basal area by species), and frequency (number of times the species occurs in the plots) (^{Curtis & McIntosh, 1951}). Then, the relative values of these attributes were calculated.

Additionally, tree ownership was recorded as publicly owned if the tree was located in open public space, including sidewalks; or privately owned when the tree was located inside the limits of a building, house, or a lot. To measure privately owned trees, residents were asked for permission to access their property and take measurements along with a brief explanation of the purpose of the work.

Data analysis

Species diversity was quantified as species richness (the total number of species recorded by plot), and diversity estimates were performed with the package iNEXT in R (^{Chao et al., 2014}; ^{Hsieh et al., 2016}). To compare tree species richness within land use types and boroughs, sample-size based rarefaction and extrapolation curves with Hill numbers of order q = 0 (species richness), q = 1 (Shannon diversity), and q = 2 (Simpson diversity) were used (^{Chao et al., 2014}; ^{Hsieh et al., 2016}). Differences in species richness were evaluated using confidence intervals (95%) (^{Chao et al., 2014}). Beta diversity analyses were performed in the iNEXT.beta3D package (^{Chao & Hu, 2024}) and was evaluated using the dissimilarity index Jaccard-type turnover, where β-diversity of 0 indicates identical species composition among plots and 1 represents completely distinct communities.

The Kruskal-Wallis non-parametric test was used to investigate differences in plot-level structural characteristics (basal area, canopy cover, number of trees, and tree height) for land use types and boroughs. When significant differences were found, the Wilcoxon rank sum test was used to identify differences between categories at p < 0.05. The Bonferroni method was used to correct for multiple comparisons. Data analyses were performed using R version 4.3.2 (^{R Core Team, 2022}).

Urban forest diversity and structure

Of the 500 originally targeted plots, 320 (64%) contained trees and 180 (36%) were without trees. In the 320 plots (12.8 ha) with trees, 1640 trees were surveyed. A total of 106 tree species, 72 genera, and 44 families were found (The complete list of species can be consulted in the supplementary material 2). The Shannon diversity index of 33.8 (± 1.1 SE), and a Simpson diversity index of 18.9 (± 0.7 SE). Overall, 30% of all the tree species were native and 70% of the species were introduced. Evergreen species represented 64% and deciduous species represented 36% of the tree species. According to their biogeographical origin, trees were mostly represented by sub-tropical species (45%), followed by temperate and tropical tree species (27% each).

The most important families in terms of tree species were Rosaceae (10 species), Pinaceae (9), Cupressaceae (6) and Myrtaceae (5). Genera represented with the most species were Pinus (9 species), followed by Prunus (5); Citrus (4); Cupressus (4); Ficus (4); and Eucalyptus (3). Tree species represented with more individuals were Shamel ash (Fraxinus uhdei) (12.4%), Mexican white cedar (Cupressus lusitanica) (9.7%), casuarina (Casuarina equisetifolia) (9.6%), weeping fig (Ficus benjamina) (8.3%), and Chinese privet (Ligustrum lucidum) (5.7%). The most abundant introduced species were casuarina (C. equisetifolia), weeping fig (F. benjamina), Chinese privet (L. lucidum), blue gum (Eucalyptus globolus) and Mediterranean cypress (Cupressus sempervirens). As for native species, Shamel ash (F. uhdei), Mexican white cedar (C. lusitanica), American sweet gum (Liquidambar styraciflua), butterfly bush (Buddleja cordata) and Bonpland willow (Salix bonplandiana) were the most frequent.

Most of the species were represented by less than five tree individuals, for instance, 24 species were represented by only one individual (e.g., Chiranthodentron pentadactylon, Juglans nigra, Ginko biloba, Robinia pseudoacacia). The most important species ranked by IVI were F. uhdei (13.3%), C. lusitanica (10.0%), C. equisetifolia (8.2%), F. benjamina (6.7%), and the palm Phoenix canariensis (4.9%). While these species accounted for 43.1% of the total importance, the remaining 106 make 56% of all species.

Trees located on public property represented 71.2% and trees on private property accounted for 28.8% of the individuals. In public property 84 species were registered of which 71.5% were non-native species and 28.5% native species; 51.2% were sub-tropical species, 25.0% temperate, and 23.8% tropical; and 66.6% were evergreen and 33.4% deciduous species. As per private property, there were a total of 70 species, of which 67.2% were non-native and 32.8% native; 47.1% were sub-tropical, 27.1% temperate, and 25.8% tropical species; and 68.5% were evergreen and 31.5% deciduous species. The total basal area in public property was 92.17 m²/ha and in private property was 36.01 m²/ha. Tree height of all trees in public property was 8.9 ± 0.1 and 7.7 ± 0.1 in private property.

The mean DBH was 25 cm (± 4.7 SE). The size distribution of trees ranged from 5 cm to 170 cm DBH, with half of the trees having DBH < 20 cm, while 32.7% had a diameter between 20 and 40 cm, 12.3% between 40 and 60 cm, and 5% > 60 cm (Fig. 2). The total basal area of the sampled trees was 128.1 m²/ha, and the average basal area per plot was 0.405 m²/ha (± 0.03 SE). The total canopy cover of the sampled trees was 8095 m² with an average canopy cover per plot was 25.3 m²/ha (± 1.4 SE). The average tree height per plot was 7.8 m (± 0.18 SE).

Bars represent standard error

Figure 2 Distribution of diameter at breast height (DBH) classes of the sampled trees across the urbanized area of Mexico City (N = 1640), mean 25 cm (± 4.7 cm). 

Urban forest variation by land use types and boroughs

Results indicated that significant differences exist in plot-level urban forest structural characteristics across land use types and boroughs (Table 1). Green areas had higher basal area, canopy cover, number of trees, and tree height than the rest of land use types; only for tree height, green areas were similar to the industrial land use. The commercial-residential land use had the lowest values for all structural variables, except for tree height (Fig. 3). Within land use types, species richness ranged from 8 to 68 in the industrial and residential land use, respectively (Supplementary material 1). A clear trend for changes in species richness within land use types was observed. The agricultural and industrial land uses were similar and registered the lowest values of species richness, while green areas and services formed a distinctive group with intermediate values of species richness. The residential land use had the highest species richness and Shannon diversity, while for the Simpson diversity both residential and services have the highest values (Fig. 4). All land use types except for agriculture, registered a higher number of non-native species than native species (Supplementary material 1).

The same letters indicate no significant difference, different letters indicate statistically significant differences resulting from Kruskal-Wallis test and Wilcoxon post-hoc tests (p < 0.05).

Figure 3 Mean values of structural characteristics of the urban forest of Mexico City by land use type. 

Figure 4 Sample-based rarefaction curves with Hill numbers of the orders q = 0 (species richness), q = 1 (Shannon diversity), and q = 2 (Simpson diversity) of tree species within land use types in the urbanized area of Mexico City (95% confidence intervals). 

The highest values of canopy cover and tree density were found in the borough of Coyoacán. Tree height was similar between boroughs and the lowest values were recorded in the boroughs Gustavo A. Madero, Iztapalapa, Magdalena Contreras, and Tláhuac. The borough Benito Juárez had the highest basal area (Fig. 5). The borough Iztacalco located in the east part of the city, had the lowest species richness (8 species) while Tlalpan found in the south had the highest (47 species) (Supplementary material 1). Differences in the species richness indicated by rarefaction curves showed one group of boroughs with lower species richness formed by Azcapotzalco, Cuajimalpa, Tláhuac, and Venustiano Carranza; another group with higher species richness was formed by Álvaro Obregón, Benito Juárez, Coyoacán, Cuauhtémoc, Gustavo A. Madero, Iztapalapa, Tlalpan, and Xochimilco, found in the center, south and east parts of the city. The borough Miguel Hidalgo separated from these former groups (Fig. 6). For Shannon diversity, both Tlalpan and Xochimilco had the highest values, and for Simpson diversity Milpa Alta and Xochimilco were the ones with the highest values (Fig. 6).

The same letters indicate no significant difference, different letters indicate statistically significant differences resulting from Kruskal-Wallis test and Wilcoxon post-hoc tests (p < 0.05).

Figure 5 Mean values of structural characteristics of the urban forest of Mexico City by borough. 

Figure 6 Sample-based rarefaction curves with Hill numbers of the orders q = 0 (species richness), q = 1 (Shannon diversity), and q = 2 (Simpson diversity) of tree species within boroughs in the urbanized area of Mexico City (95% confidence intervals). 

Taxonomic beta diversity (dissimilarity) was slightly higher for boroughs (β = 0.63) in comparison to land use types (β = 0.57). Agricultural land use is different in its species composition than the remaining of land uses, whereas the commercial-residential is similar to green areas, residential and services, and the industrial land use is only similar to green areas (Table 2). Within boroughs, Cuajimalpa, Iztacalco, Milpa Alta, Magdalena Contreras, Tlahuac, and Venustiano Carranza are different between them and than the rest of boroughs, and these are all located farther from the central areas of the city (Table 3).

Urban forest variation across land use types and boroughs

As expected, land use types and boroughs were ecologically heterogeneous units as indicated by their variations in structure and diversity, which reinforces the idea that urban forest planning needs to be targeted considering the site’s specific conditions. The main differences in structural characteristics between land-use types were found in the green areas. For all structural variables, green areas were significantly higher than the rest of land uses, which can be partially explained by the space availability, the age of the green area, and the management as compared to other land uses. Tree diversity was higher in residential areas, and this is similar to the findings of ^{Ortega-Álvarez et al. (2011)}, in their analysis of tree diversity across an urbanized gradient in central and southwestern Mexico City, finding a positive asociation between species richness and residential land-use and higher species richness at intermediate developed lands. Both residential and commercial-residential registered higher values of Shannon diversity, indicating more common species within these land uses, whereas for the Simpson diversity the residential and services were the ones with higher values, which indicates more dominant species, individuals between species are more evenly distributed within the residential, commercial-residential and services land use types.

The findings from the present study indicated that the structural variables evaluated here were consistently similar within boroughs located in the center, south, and west parts of the city, while boroughs in the east part formed similar groups with lower values for structural variables. Similarly, boroughs in the east part also had lower species richness, suggesting higher urban forest planning needs in Venustiano Carranza, Iztacalco, Iztapalapa, and Tláhuac. These areas are dominated by hard surfaces (^{Toledo-Garibaldi et al., 2024}), thus reducing the availability of planting spaces and limiting urban forest planning. Demineralization of paved surfaces is a limited option in Mexico City due to the intensity of use of the public space and this strategy could accentuate existing space conflicts. Nonetheless, some wide sidewalks, or spaces in parks, schools, or institutional buildings could be transformed into soft surfaces to plant trees with small to medium final dimensions. Other green infrastructure can be implemented such as green walls, climbing plants, and green roofs. However, there are areas without trees and with no possibility of increasing greenery due to their specific conditions (e.g., industrial sites). Borughs located in the south (Tlalpan, Xochimilco and Milpa Alta) had the highest Shannon and Simpson diversities, showing their ecological importance, and the need to diversify boroughs towards the northern part of the city.

‌The information obtained in this study can serve as a first step toward the development of a comprehensive urban forest management plan (UFMP). While in North America numerous cities count with UFMP (e.g. Montreal, Ottawa, Quebec City, Toronto, Vancouver, New York, Oakland, San Francisco, Seattle), Mexico lacks compre-hensive management plans at the city scale, and accounting for urban forest variations that highlight the urban heterogeneity is necessary for a better planning and management. A UFMP based on tree inventory data is essential for protecting and planning urban forests and is an effective tool that gives planners, citizens, and government information and recommendations needed to manage public trees. Effective management can help maximize the environmental services of urban trees, increasing the environmental quality of cities and the well-being of city dwellers. By developing and implementing a city-scale management plan, Mexico City would start a long-term commitment to manage, protect, and expand the urban forest, ensuring the provision of environmental services to the city residents.