Study area

The MFNNP is located in the municipality of Texcoco, Estado de México, and is considered a peri-urban forest (Figure 1). According to the perimeter polygon, the total area covers 49.2 ha (^{Secretaría del Medio Ambiente y Recursos Naturales & Comisión Nacional de Áreas Naturales Protegidas [SEMARNAT & CONANP], 2017}). The park is located at an altitude of up to 2 300 m and the dominant climate is temperate with an average annual temperature of 14 °C and precipitation of 620 mm (^{Herrera-Llampallas et al., 2018}). This park stands out in the area due to its historical, cultural, and recreational relevance for visitors to the Valley of Mexico.

Figure 1 Mapping of Molino de Flores Netzahualcóyotl National Park, Estado de México and location of sampling sites. 

Sampling design

For tree measurements in the MFNNP, the inventory was determined using stratified random sampling. Eleven management zones (strata) were determined based on the official perimeter polygon (^{SEMARNAT & CONANP, 2017}), zoning carried out by ^{Herrera-Llampallas et al. (2018}), field surveys, and the information provided by the park management. In this study, an intensity of 5 % of the park's surface was used, representing a total of 62 sampling sites. These were randomly selected using a 25 x 25 m grid of points and assigned proportionally to the size of each management zone.

The shape of the sampling sites was circular with an area of 400 m², where all trees larger than 7.5 cm diameter at breast height were recorded. The variables recorded for each tree were: species, diameter at breast height (1.30 m in height from the base), total height and projection of crown diameter measured from north to south and from east to west. Subsequently, in the laboratory, the average crown diameter was calculated using the measurements of the two projections of each tree, and the crown cover (CC, m²) of all living individuals, which is defined as the proportion of the surface area that is covered by the vertical projection of the tree crown (^{Korhonen et al., 2006}). In addition, the percentage of live crown and crown density were recorded according to the sampling procedure of the Comisión Nacional Forestal (^{CONAFOR, 2017}); these variables were subsequently used when calculating the urban importance value index (UIVI) (^{Saavedra-Romero et al., 2019}).

Data analysis

A Microsoft Excel® database was used to characterize the forest structure with the formulas proposed by ^{Montesinos et al. (2009)}, which allowed the population estimation of each management zone for the following variables: tree density, total canopy cover, average diameter at breast height and average total height. Diversity was determined using the Margalef richness index (D_MG), Simpson's diversity index (1-D) (^{Magurran, 2004}) and the 10-20-30 diversity rule of ^{Santamour (1990}). This rule indicates that, to ensure greater protection against pest and disease attacks, individuals of the same species should not exceed 10 % of the total number of trees, species of the same genus should not exceed 20 % and trees of the same family should not exceed 30 % of the total number of trees.

A cluster analysis was performed using species composition and abundance data to identify management zones that share species and facilitate management decisions. This analysis was carried out using Ward's method, Bray-Curtis distance and a 0.4 cut-off line; in addition, the cophenetic correlation coefficient was calculated. The variables diameter at breast height, total height, Margalef richness and Simpson's diversity index (1-D) were subjected to a non-parametric Kruskal-Wallis analysis of variance to determine if there are significant differences between each of the park's management zones. It was decided to use non-parametric statistics because the variables were not normally distributed according to the Shapiro-Wilks test. These analyses were carried out with the statistical software InfoStat 2020 (^{Di Rienzo et al., 2020}) with a significance level of 5 %. Finally, the importance of each tree species, horizontally, vertically and three-dimensionally, was hierarchized with the urban importance value index (UIVI) proposed by ^{Saavedra-Romero et al. (2019}).

Composition

We measured 404 trees in all strata, with 89.1 % being live trees and 10.89 % being standing dead individuals. In forests with high recreational value, the presence of standing dead trees can represent a potential threat to visitors (^{Stereńczak et al., 2017}). Because of this, it is advisable to diagnose the risk of tree fall to be able to mitigate or reduce it, even if standing dead trees represent an element of wildlife habitat.

According to Table 1, the living tree population consists of 17 families and 22 species. The most frequent species are Eucalyptus camaldulensis Dehnh., Schinus molle L. and Casuarina equisetifolia L. Despite being a natural-protected area (NPA), the MFNNP has an arboreal composition more similar to an urban forest. For example, in the core zone of the NPA Reserva de la Biosfera Mariposa Monarca, a composition of 49 native tree species has been reported (^{Cornejo-Tenorio & Ibarra-Manríquez, 2017}), while, in the MFNNP, of the 22 tree species recorded, 10 are introduced and represent 62 % of the total population sampled. This is because, in peri-urban ecosystems, the loss of components and ecological functioning of natural ecosystems is common, as well as the increase of elements managed by humans (^{Tlapa et al., 2020}).

In urban and peri-urban forests, the presence of many exotic species is frequent (^{Templeton et al., 2019}). For example, in five cities in central Chile, exotic species represent 82 % of the total (^{Santilli et al., 2018}) and in 15 Italian cities, 277 tree species were recorded, of which 68 % are exotic species, mainly originating from Asia, America and Australia (^{Bartoli et al., 2021}). In Mexico, exotic species can represent a large proportion of the tree population in urban areas; for example, in the Bosque San Juan de Aragón, the tree population is mainly composed of C. equisetifolia, E. camaldulensis andGrevillea robusta A. Cunn. ex. R. Br.,which represent 53 % of the population (^{Saavedra-Romero et al., 2019}). This situation is similar to that presented in the MFNNP, where the exotic species C. equisetifolia, E. camaldulensis and S. molle represent 55.4 % of the sampled population (Table 1).

The cluster analysis identified four groups of management zones with a cophenetic coefficient of 0.87 (Figure 2), which indicates a good fit in the grouping (^{Zambrano et al., 2003}). Each group includes management zones that are similar to each other in species composition. Group 1 includes the Chapel and the Recreation area; the Coxcacuaco River flows through these zones, which is why there is more water available, and they are the only two zones where Taxodium mucronatum Ten was recorded. Group 2 includes the Forest path and the Hacienda. The first zone is considered the one with the best conservation of natural resources, so it can be used as a reference to improve the other zones of the park. Group 3 includes the Restoration area 1, 2 and 3, the Garden and the Shopping area, where E. camaldulensis dominates structurally. This group can be recognized as a high priority zone for maintenance and diversification of tree species, even the conservation status of natural resources in Restoration area 1, 2 and 3 has been diagnosed as deteriorated (^{Herrera-Llampallas et al., 2018}). Group 4 includes the Nursery and the Spot. In these areas, the main species is S. molle, which is dioecious and naturalized and can reduce the germination of native species by allelopathic effects (^{Avendaño-González et al., 2016}), a situation that should be taken into account when choosing species for reforestation.

Figure 2 Dendrogram of management zone grouping according to tree species composition in Molino de Flores Netzahualcóyotl National Park, Estado de México, using Ward's method and Bray-Curtis distance. 

Tree structure

Table 2 shows trees’ structural characteristics at the MFNNP. A total population of 7 983 ± 806 trees and a density of 163 trees∙ha^-1 was estimated. In urban forests, it is common to find low density values compared to natural forests (^{Nowak & Greenfield, 2018}). Common planting density values in urban areas vary between 100 and 200 trees∙ha^-1 (^{Pincetl et al., 2013}). For example, the density in four urban parks in Texcoco City is 167 trees∙ha^-1 (^{Martínez-Trinidad et al., 2021}), while in the urban forest of Alburrá Valley, Colombia, is 133 trees∙ha^-1 (^{Arroyave-Maya et al., 2019}). It is known that visitors to urban green areas negatively perceive very low or very high tree densities (^{Campagnaro et al., 2019}). However, although it is complex to determine an optimal tree density, some park visitors in the United States prefer sites with densities between 150 and 160 trees∙ha^-1 (^{Schroeder & Green, 1985}); while people from Sweden prefer to visit recreational forests with densities of 300 trees∙ha^-1 (Pettuco et al., 2018). Although tree density preferences may vary and change over time, the above values can be taken as a reference in setting initial reforestation objectives in areas of the MFNNP with low tree density such as on the Spot, the Nursery and the Garden (Table 2).

Population structural variables correspond to an average diameter at breast height of 26.56 ± 2.33 cm and an average total height of 10.52 ± 0.6 m. Kruskal-Wallis analysis indicates that these variables differ significantly between management zones (P < 0.001; Table 2). Diameter at breast height is mainly distributed between 7.5 and 22.4 cm, measurements that indicate a population of slender trunks with a typical inverted "J" distribution (Figure 3A). This distribution is common in several urban forests; for example, there is evidence that, in 32 U.S. cities, the diameter distribution of the tree stand is mainly of the juvenile type, where more than 40 % of the tree population is concentrated in diameters <15 cm (^{Morgenroth et al., 2020}). The inverted J structure indicates that the population is growing and productive towards more advanced stages (^{Imaña et al., 2011}). This condition can be explained by the use of small trees in the reforestation programs of the park.

Figure 3 Diameter distribution (A) and total height classes (B) of the trees sampled in Molino de Flores Netzahualcóyotl National Park, Estado de México. 

Regarding total height, the Shopping and Recreation area had the highest average values with 15.69 ± 7.9 and 14.77 ± 6.8 m, respectively. These areas have a significant difference with the Spot (Table 2), where the lowest tree height was reported with 6.67 ± 4.2 m and where S. molle, the most abundant species, reaches heights of 4 to 15 m. In contrast, E. camaldulensis, F. uhdei (Wenz.) Lingelsh. and Hesperocyparis lusitanica (Mill.) Bartel, large species that reach up to 30 m in height, are found in the Shopping area (^{Hirsch et al., 2020}). In general terms, total tree height had a positive asymmetric distribution with an asymmetry coefficient of 1.08; values were mainly concentrated between 7.5 and 12.4 m (Figure 3B). This distribution is similar to the urban tree stand in the city of Linares, Nuevo León, where the total height is mainly distributed between 6.4 and 9.6 m (^{Leal et al., 2018}).

Another indicator of tree structure is canopy cover (CC). The MFNNP has an average CC of 4 819.88 m²∙ha^-1, which is equivalent to 48.2 % of the park's total area (Table 2). Compared to urban and peri-urban forests, the total CC of the MFNNP is lower; for example, the gallery forest cover within the urban area of the city of Linares is 25 307.3 m²∙ha^-1 (^{Canizales-Velázquez et al., 2021}). However, there are areas of the MFNNP with high values, such as the Recreation area, the Hacienda, and the Forest path with covers of 14 746.9, 11 247.44 and 10 621.88 m²∙ha^-1, respectively. This cover is related to a higher density of trees (Table 2); in addition, the presence of larger species contributes to the development of a larger crown diameter and, therefore, increases the cover that in these areas is higher than that reported for the 2nd section of Bosque de Chapultepec with 8 482.7 m²∙ha^-1 (^{Benavides & Fernández, 2012}). In contrast to the previous areas, the Garden, the Nursery and the Spot have low CC (Table 2). This is important because these areas are the largest within the park and the CC barely represents 12.7 %, 17.8 % and 28.9 %, respectively. The Garden area has areas with compacted soil and is classified as moderately conserved; in the Nursery area there is little organic matter in the soil, and the Spot is a deteriorated area concerning natural resources, since pest problems have been reported in most of the tree species in this area (^{Herrera-Llampallas et al., 2018}). Increasing CC in these areas can provide protection and recovery to the soil, due to reduced erosion from rainfall effects and increased organic matter content.

Diversity and Urban Importance Value Index (UIVI)

Table 3 reports the richness and diversity indices per MFNNP management area. According to the Margalef index, species richness by management zone ranges from 0.06 to 1.67 and varies significantly among management zones according to the Kruskal-Wallis analysis (P = 0.0154). The highest richness was recorded in the Recreation area with 11 species and 69 trees measured. ^{Alanís et al. (2020}) indicate that values lower than 2 suggest low richness. Regarding Simpson's index, the Kruskal-Wallis test detected significant differences between management areas (P = 0.0164) (Table 3). The lowest value corresponds to the Garden area with three tree species and 16 trees; on the contrary, the area with the highest diversity is the Recreation area with a value of 0.61 ± 0.34. Although Simpson's index and other diversity indices are sensitive to sample size (^{He & Hu, 2005}; ^{Xu et al., 2020}), which limits the comparison with other areas, there is evidence that it is common to find high tree diversity in urban areas, which can vary according to location and is generally higher than in peri-urban areas (^{Blood et al., 2016}) such as MFNNP. For example, in the city of Santiago, Chile, urban forest tree diversity has remained stable for 12 years with Simpson's index (1-D) values above 0.9 (^{Hernández & Villaseñor, 2018}). In Mexico, the tree diversity of four forest parks in the city of Texcoco has a Simpson index equal to 0.73 (^{Martínez-Trinidad et al., 2021}), which is higher than any of the MFNNP management areas.

In addition to the low richness and diversity values, the species composition of the MFNNP fails to meet ^{Santamour's (1990}) 10-20-30 rule, because three species (E. camaldulensis, S. molle and C. equisetifolia) each represents more than 10 % of the total number of trees sampled and the genus Eucalyptus represents more than 20 % (Table 1); however, no family exceeded 30 % of the total sampled trees. This indicates the existence of closely related and similar species that respond to the same stress factors and, therefore, susceptibility to pest and disease attack could increase (^{Paquette et al., 2021}). Considering this scenario, it is necessary to increase tree diversity, mainly in the Garden, Restoration area 1 and the Spot in order to reduce the risks of pest and disease attack (^{Herrera-Llampallas et al., 2018}). In addition, the increase in diversity and species composition provides greater resistance and resilience that allows to obtain greater benefits by trying to ensure the supply of ecosystem services (^{Morgenroth et al., 2016}).

Regarding the Urban Importance Value Index (UIVI), Table 4 shows that E. camaldulensis is the dominant species with 38.53 %, followed by S. molle (UIVI = 13.89 %) and F. uhdei (UIVI = 10.62 %). These three species have the highest ecological importance within the MFNNP calculated from dominance, frequency, height, crown volume and surface area; however, this does not necessarily mean that they should be used or promoted in future reforestation programs, because they only reflect the current dominance in the urban forest structure (^{Koricho et al., 2020}). The UIVI is an index that considers horizontal, vertical and three-dimensional parameters that allow the identification of tree species with high biomass levels that can provide greater benefits to the population (^{Saavedra-Romero et al., 2019}). However, there is a few research of their use in other urban green areas. The UIVI results within the MFNNP are similar to those reported for the Bosque San Juan de Aragón in Mexico City, where C. equisetifolia is the dominant species and E. camaldulensis is the co-dominant one with values of 31.98 % and 19.96 %, respectively (Saavedra-Romero et al., 2019). In the MFNNP, although the greatest importance is represented by two exotic species, the presence of F. uhdei, a species native to Mexico, is noteworthy, possibly due to the sensitivity of the UIVI to high biomass levels.

In urban and peri-urban areas, it is common that the most important species, as measured by conventional importance value indices, are exotic species. For example, in an area in Guadalajara, Jalisco, the species with the greatest ecological importance is F. benjamina with an importance value index (IVI) of 19.34 % (^{Hernández et al., 2022}). In another case, for the cities of Linares and Montemorelos, Nuevo León, the species with the highest ecological importance was Fraxinus americana L., an introduced taxon with IVI of 30.91 % and 53.82 % for each city, respectively (^{Canizales et al., 2020}; ^{Leal et al., 2018}). These results evidence the importance of exotic species within urban and peri-urban forests because they can generate important ecosystem and economically valuable benefits (^{Riley et al., 2018}), whether measured with conventional IVI or the new UIVI. While the use of native species is recommended as having greater adaptive and growth advantages, exotic species cannot be completely excluded from the urban forest since this action may jeopardize the resilience of urban and peri-urban ecosystems (^{Sjöman et al., 2016}).