Materials and Methods

Data records. Specimen data of Hampea were obtained from the following herbaria: Paul Fryxell herbarium (pf) now part of the New York Botanical Garden herbarium (NY); Herbario Nacional de México, México D.F. (MEXU); Instituto de Ecología, Xalapa, Veracruz, México (XAL); El Colegio de la Frontera Sur, San Cristóbal, Chiapas, México (CH); and el Herbario de la Universidad Autónoma de Querétaro "Jerzy Rzedowski," Querétaro, México (QMEX). Data from these herbaria were cross-referenced with that of the online national Mexican data base of the National Commission for the Knowledge and Use of Biodiversity (^{CONABIO 2016}). Observations and collection data by the authors conducted during the last 25 years were also incorporated into the data base. Distribution data was evaluated and data points selected based the author’s validation of species identity and quality of locality data.

A total of 218 records were used to generate the niche models, the total records used per species were: H. breedlovei Fryxell, 9; H. integerrima Schltdl., 36; H. latifolia Standl., 8; H. longipes Miranda, 12; H. mexicana Fryxell, 12; H. montebellensis Fryxell, 8; H. nutricia Fryxell, 47; H. rovirosae Standl., 16; H. stipitata S. Watson, 30; H. tomentosa (C.Presl.) Standl.,10; and H trilobata Standl. 30.

Environmental variable selection for modeling of potential distribution. We considered 20 variables to generate the ecological niche models: 19 climatic variables and a variable for soil type. The climatic variables used were specific to Mexico using information from climatic stations for the period 1910 to 2009, in turn represent annual averages, seasonality and extreme factors potentially correlated with the presence of the species (^{Cuervo-Robayo et al. 2013}). All variables were used in raster format with a 1 km² resolution. The soil map used was “Conjunto de Datos Vectoriales Edafológico”, scale of 1:250,000, Serie II (^{INEGI 2016})

To insure independence of variables (^{Tsoar et al. 2007}), a selection of the most important variables from the 20 initial set was performed, which in turn resulted in lowered inter-variable correlations based on a Pearson correlation matrix for the variables using SPSS 20 (^{IBM 2011}). A coefficient of (r) > 0.7 in the resulting matrix was selected to identify the most correlated variables, and variables that showed the lowest correlation coefficients (r) < 0.7, were used to generate the niche models for the species (Table 1).

Algorithm parameters. Potential distribution and environmental suitability predictions for species were performed with the Maximum Entropy Algorithm program (Maxent^®; Version 3.3.3k) (^{Phillips et al. 2006}). Algorithm widely used to understand the relationship between environmental variables and the presence of species (^{Elith et al. 2011}).

The program's default settings used were: regularization multiplier in one, convergence threshold in 10^-5 and the maximum number of background points in 10,000 (which serve as pseudo-absences for Maxent analyzes) was maintained. A maximum of 1,000 iterations (^{Phillips et al. 2006}) were chosen and the selection of the Extrapolate and Clamping options, which are used to generate future predictions, was omitted.

Hampea integerrima, H. rovirosae, H. trilobata, H. nutricia and H. stipitata, had between 16 and 47 presence records, with which 10 replicates were generated per species, using the "random seed" and "subsample" options and a random test percentage of 20 %, used in similar works (^{Lodoño-Murcia & Sánchez-Cordero 2011}). This allowed the program to use that percentage to generate a Receiver Operating Characteristic (ROC) curve for each model, with an associated value of AUC (area under the curve) for both sets of data, with 80 % to generate the model and 20 % to test it (^{Phillips & Dudik 2008}). Models with an AUC above 0.75 were considered to be potentially good predictive models (^{Elith et al. 2002}, ^{Phillips et al. 2006}). Based on the above criteria, only one model of the 10 replicates generated by species was chosen with those with an AUC higher than 0.85 (^{Marmion et al. 2009}).

For the species with fewer than 15 records, H. breedlovei, H. latifolia, H. longipes, H. mexicana, H. montebellensis and H. tomentosa, all records available for each species were used to generate each of the models, without using 20 % of them as a test within the program and without generating replicas. In these cases, the criteria to make ecological niche predictions only required presence records and thus, was less sensitive to the record numbers (^{Hernández et al. 2006}), and in some cases, five records was sufficient to generate a useful model (^{Pearson et al. 2007}).

The ecological niche spatial predictions generated with Maxent have a scale of 0 (not suitable) to 1 (highly adequate), this probability influences the distribution of the species (^{Phillips et al. 2006}). In all models, the logistic output format and the ACII format for the predictions were selected which allowed for a simpler interpretation of the model. The Jackknife analysis was also used to examine the contribution of each environmental variable (^{Phillips & Dudik 2008})

Threshold cut selection. Although there are different alternatives to determine optimal threshold selection to transform the resulting probability into absence-presence data on a map (^{Freeman & Moisen 2008}), there are no set rules to determine an optimal threshold (^{Hu & Jiang 2010}). Due to characteristics of the collection records and based on recommendations of ^{Freeman & Moisen (2008)}, a threshold corresponding to the 10th percentile was chosen, removing from the final projection 10 % of the distribution predicted by Maxent for the final models of each species.

Determination of the category of risk. The conservation status of species of Hampea in Mexico was evaluated using criteria given in the Methods of the Evaluation of Risk of Extinction of plants (MER-Plantas) of the Mexican Ministry of Environment and Natural Resources (NOM-059-SEMARNAT-2010,^{SEMARNAT 2010}). The four criteria of this method of the evaluation include: Criteria A, characteristics of the geographic distribution; Criteria B, habitat characteristics; Criteria C, intrinsic biological vulnerability, and Criteria D, impact of human activity. The details for evaluation of these criteria and the assignation of points are given in Nom-059-SEMARNAT-2010-Anexo Normativo (^{SEMARNAT 2010}). The amplitude of distribution (Criteria A.1) was calculated using the area of the potential distribution of each species calculated in the present work, in proportion to the total surface area of Mexico. The final variables used for the analysis of distributions were based on the independence of all variables with only those with a Pearson correlation of r < 0.8, used in the final analysis. The number of populations (Criteria A.2) was determined counting separate, non-overlapping data points (3 mm) discernable on a map of 1: 4,000,000 (^{SEMARNAT 2010}). The number of biogeographic provinces and vegetation types that species occurred was determined by mapping data points on the map of the biotic provinces (^{CONABIO 2016}) and vegetation types (^{CONABIO 2016}) as given in the MER Criteria (^{SEMARNAT 2010}) on a scale of 1:4,000,000.

Results

Table 2 presents the evaluation of the four criteria of MER-Plantas and point system for all species of Hampea in Mexico. Based on the evaluation of the criteria and the assigned points, each species was categorized within a category of risk of extinction (Table 2). The species, H. montebellensis, presently listed as “Endangered” in the NOM-059-SEMARNAT-2010 (^{SEMARNAT 2010}), as well as, H. breedlovei and H. longipes were categorized as “At Risk of Extinction” under the MER-Plantas criteria. The species, H. latifolia was categorized as “Threatened” and H. stipitata, H. mexicana, and H. tomentosa were considered as “Subject to Special Protection.” The species, H. nutricia, H. intergerrima, H. trilobata and H. rovirosae were not categorized in a threatened category and considered of least concern for conservation measures.

Discussion

Criteria A. Characteristics of the geographic distribution. Hampea breedlovei had the smallest area of distribution, calculated as the potential distribution, of analyzed species of Hampea, and represented 0.02 % of the Mexican territory (Criteria A.1 = 3; Figure 3, upper left). Initially, this species was only known from the type locality when described by ^{Fryxell in 1977} which was listed as: Mexico: Chiapas: Municipality of Rayon, in the Selva Negra 10 km above Rayon Mezcalapa along road to Jitotol . Although 10 separate collection sites appear in the Bios CONABIO data base (^{CONABIO 2016}) within 15 km of the type collection site, some of these points apparently represent the same samples that were assigned different coordinates. Only two sites are not within the vicinity of the type collection: the first is one specimen 12 km S of Tila (Municipality: Tila) above Petalcingo, Chiapas at 1,700 m (Figure 2) and the second reported is from Veracruz, municipality of Acajete, west of Xalapa. This latter record is almost certainly a misidentification of H. integerrima given that this latter species is common in the western area of Xalapa (Figure 1, personal observation, RWJ).

Figure 3 Potential distribution of six species of Hampea in Mexico: Upper left, Hampea breedlovei Fryxell; upper right, Hampea latifolia Standley; middle left, Hampea montebellensis Fryxell; middle right, Hampea longipes Miranda; bottom left, Hampea mexicana Fryxell; bottom right, Hampea integerrima Schlechtendal. 

Hampea latifolia was the species with the second smallest distribution representing 0.14 % of the Mexican territory (Criteria A.1 = 3, Figure 3, upper right). This species is an endemic of the Soconusco region of southeastern Chiapas and adjacent departments of Retalhuleu and Suchitepéquez in southwestern Guatemala. Within Mexico, only four specimens were found in the National Autonomous University of Mexico (UNAM) herbarium and only one collected after 1980. In Chiapas, ^{Jones (1994)} observed five trees in forest fragments and in cacao plantations near the town of Palo Seco in the municipality of Tapachula, and seven trees from 11 to 19 km north the municipal seat of Mapastepec, in the foothills of the Cerro El Triunfo, near sites recorded for the MEXU herbarium collections.

Hampea montebellensis is another species with very limited distribution occurring within 0.15 % of the Mexican territory (Criteria A.1 = 3, Figure 3, middle left). The species has only been found from central and eastern Chiapas from 900 to 1,500 m in moist forests (^{Fryxell 1988}). Of the 44 specimen records in the CONABIO data base (^{CONABIO 2016}), only four of the localities were not from within or in the vicinity of the Lagunas de Montebello National Park.

Hampea longipes had a potential distribution that covered 0.99 % of the Mexican territory (Criteria A.1 = 3, Figure 3, middle right). Individuals of this species have been well collected in the state of Chiapas and Oaxaca, with 95 records in the CONABIO data base (^{CONABIO 2016}). Recorded collection of this species indicates that it is endemic to the highlands (“Los Altos”) of northern Chiapas and also found at elevations above 1,000 m in the adjacent Chimalapas region of eastern Oaxaca. Although, the potential distribution suggests that this species may occur in the Sierra Madre de Chiapas, no herbarium specimens from the region were found and flora studies (El Triúnfo) do not mention species of Hampea (^{Martínez-Meléndez et al. 2008}, ^{Pérez-Farrera et al. 2012}). Specimens of Hampea in the central highland region of Oaxaca have been identified as both H. longipes and H. intergerrima, but no fruiting specimens were observed in herbaria to verify species identity from this state.

The potential distribution of H. mexicana presented here (Figure 3, lower left) represents 1.74 % of the Mexican territory (A1 = 2). This distribution is within the central Depression of Chiapas, extending into the Chimalapa region of eastern Oaxaca in seasonally dry forests, from 800 to 1,000 m. However, there are no records of the species from central Oaxaca and Guerrero as indicated by the potential distribution. Also specimens misidentified as H. stipitata were frequently found in various herbaria, especially from the Lacandon region giving the erroneous impression that the distribution of H. mexicana encompasses northern and northeastern Chiapas.

The distribution of H. integerrima in the southern Sierra Madre Oriental (SMO) and tropical portions of the Neovolcanic axis in Veracruz within populated areas near the montane forests around Córdoba and Xalapa, Veracruz (Figure 3, lower right) has resulted in this species being one of the most collected of the genus. The species has not been reported much further north than 20^o latitude, although the potential distribution suggests that suitable humid montane habitat extends north to Tamaulipas and covers 0.38 % of the national territory (Criteria A.1 = 3). Additionally, the potential distribution extends from the SMO through the Neovolcanic Axis into Oaxaca, but there are only a few specimens from this state in herbaria and all samples seen were vegetative, making identification difficult. Trees were observed in the field in Veracruz by the authors including on the western edge of Xalapa, near the Cascada de Texolo, the slopes of Cerro Orizaba, and in the outskirts of Córdoba.

Hampea stipitata is relatively well represented in herbaria (^{CONABIO 2016}) although its distribution covers only 0.33 % of Mexico (Critiera A1 = 3), suggesting that the species is common in portions of its distribution. This species is restricted to southern Mexico in lowland tropical forests of southeastern Veracruz, northeastern Oaxaca into the Lacandón region of Chiapas (Figure 4, upper left) and extends into the Petén region of Guatemala. The distribution of the species also extends south from the lowland areas of Lacandón region to the west central region of Chiapas, reaching forests at altitudes close to 1,000 m in the municipalities of Berriozábal and Ocozocoautla.

Figure 4 Potential distribution of five species of Hampea in Mexico: upper left, Hampea stipitata S. Watson, upper right, Hampea trilobata Standley; middle left, Hampea rovirosae Standley, middle right, Hampea tomentosa (Presl.) Standley, lower center, Hampea nutricia Fryxell. 

Hampea trilobata is essentially a Yucatán peninsula endemic (Figure 4, upper right). There are more specimens of this species in herbaria of Mexico than all other species, an indication of its relative high abundance, especially considering it is not a showy tree. Its distribution covers 7.00 % of the Mexican territory (Critiera A1 = 1)

Hampea rovirosae is the only species with a marked disjunct distribution (Figure 4, middle left). It is a lowland species being found in humid tropical forests of extreme southern Veracruz, throughout Tabasco, northwestern Chiapas and southwestern Campeche but also reported from the Soconusco region of the Pacific coast of Chiapas near Puerto Madero (^{Bodegas-Valera et al. 1977}). The species covers 1.20 % of the Mexican territory (Criteria A.1 = 1).

The potential distribution of H. tomentosa is one of the more extensive that we calculated for the genus, covering 3.18 % of the Mexican territory (Criteria A.1 = 2, Figure 4, middle right). However, the species is poorly represented in herbaria with only 36 recorded specimens in the CONABIO data base (^{CONABIO 2016}), and was difficult to find in the field (RWJ, personal observation), suggesting low population densities.

The distribution of H. nutricia is one of the largest for the genus, ranging from northern Chiapas and Tabasco, eastern Puebla, and along the gulf coast of Veracruz, north almost to 22^o latitude (Figure 4, lower center) and covers 7.94 % of Mexico (Criteria A.1 = 1). It is mainly a lowland species but is also found in lower montane rain forests below 1,000 m (^{Fryxell 1988}).

The number of populations and their presence in biotic provinces constitute Criteria A.2 and A.3, respectively. Two species had only one to three recorded populations (A.2.a = 3): H. breedlovei (2 populations) and H. montebellensis (2). Four species had four to eight populations (A.2.b = 2): and included: H. longipes (7), H. latifolia (4), H. mexicana (5), and H. tomentosa (7). Species with nine to 25 populations (A.2.c = 1): were: H. intergerrima, H. stipitata,, and H. trilobata. The only species with greater than 26 populations (A.2.d = 0) was Hampea nutricia. All species were found in two or three biogeographic provinces which places them all in Criteria A.3 = 2.

None of the populations of species of Hampea in Mexico have what can be considered peripheral or extralimital distributions (Criteria A.4). Only five of the species, H. latifolia, H. mexicana, H. trilobata, H. longipes, and H. stipitata are not endemic to Mexico, and although these species have distributions that extend into Central America, a significant proportion of the populations, as suggested by areas of distribution and specimen records, are in Mexico (^{Fryxell 1979}, ¹⁹⁸⁸).

Criteria B. Characteristics of the Habitat. Species of Hampea in Mexico were found from one to five different vegetation types (Criteria B1). Hampea breedlovei was only recorded from cloud forests of northern Chiapas (B.1.a = 3). Species recorded from two vegetation types were H. montebellensis, H. longipes, H. stipitata, H. latifolia and H. rovirosae (B.1.b = 2). Those found in three vegetation types (B.1.c = 1) were, H. trilobata, H. mexicana, and H. tomentosa. Hampea intergerrima and H. nutricia were reported from four and five vegetation types, respectively.

Several species of Hampea were considered to be dependent on primary habitat to sustain populations (Criteria B3) based on collection records and observations in the field. These species were: H. breedlovei, H. longipes, H. mexicana, H. montebellensis, H. stipitata, and H. latifolia. For H. breedlovei, H. longipes, and H. montebellensis, these species are restricted to humid, montane habitats and no observations were reported or observed of individuals outside such habitat. For the primarily lowland or mid-altitude species, H. stipitata, H. mexicana and H. latifolia, individuals were invariably found associated with relatively undisturbed lowland tropical forests, although at times within small fragments. Never were individuals of these species observed in recently disturbed vegetation or in open areas as is frequently found with H. nutricia, H. trilobata,or H. rovirosae.

Species of the genus Hampea are relatively restricted in their range of altitude in Mexico (Critieria B.5). However, only H. breedlovei was considered to be restricted to an interval of less than 200 m in altitude (B.5 = 3), based on collection records and data given by ^{Fryxell (1988)}. Reported altitudinal ranges (^{Fryxell 1988}) and field observations indicated that H. integerrima, H. longipes, H. mexicana, H. trilobata, and H. rovirosae, although each were found at different altitudinal zones, are limited to an altitudinal range from 200 to 500 m (B.5 = 2). Hampea montebellensis, H. nutricia, H. stipitata, H. latifolia and H. tomentosa had the greatest ranges in altitude, from 500 to 1,000 m (B.5 = 1).

Two of the MER-Plantas habitat criteria did not apply to species of Hampea in Mexico. None of the species can be considered to have a highly specialized habitat (Criteria B.2). Likewise, there was no evidence that a species required a particular regime of disturbance for population maintenance (Criteria B.4).

Criteria C. Intrinsic biological vulnerability.- Quantitative field data for the demographic characteristics (Criteria C.1) of species of Hampea is lacking. However, it was considered possible to calculate population densities estimates for H. breedlovei (C.1.1) because of its highly reduced area of distribution. The calculated population density of H. breedlovei required to surpass criteria C.1.1.b (5,000 individuals) is 13.7/km², or one tree per 7 ha, based on the predicted area of distribution. However, only 12 collections are reported for this species. Eleven of these are reported from the Selva Negra region above Rincón Chamula which has been highly altered and progressively fragmented with an almost complete reduction in forested area (^{Ramı́rez-Marcial et al. 2001}, ^{Cayuela et al. 2006}). Only one other collection record exists outside the Selva Negra area from the isolated Cerro Ahk'ubal Nab, above Petalcingo (Figure 2), which is a forested area of less than 15 km². Given the small area of distribution, the reduced number of collections and the present loss of habitat, it is highly improbable that there presently is a population density of 13.7 trees/km² of H. breedlovei in the northern highlands of Chiapas. Thus, we conservatively categorized Criteria C1 as less than 5,000 individuals (C.1.b = 2). Although other species probably have reduced populations as well, there was insufficient data to estimate population sizes (C.1.b = 0).

Other demographic characteristics (C.1) were as follows. Recruitment of seedlings were observed in most species with no data for the remainder, thus for all, C.1.2 = 0. There is also no evidence of factors involved with density dependent reproduction (C.1.3a) in Hampea, nor is there evidence of clonality (C.1.3.b = 1) (personal observation, RWJ). All species except H. rovirosae are dioecious (^{Fryxell 1979}), thus C.1.3.e = 1 for all species except H. rovirosae (0). There is no evidence of high variation in fecundity, C.1.3.d = 0 for all species. Fruiting in Hampea is not characterized by few propagules (C.1.3.g) and is often prodigious for large trees (^{Fryxell 1979}, personal observation, RWJ).

There is evidence of population decline (C.1.3) for several species. Significant destruction of the type habitat of H. breedlovei is discussed previously. The destruction of the type habitat H. montebellensis within the Montebello National Park due to fires and deforestation habitat is well documented (^{Román-Cuesta & Martínez-Vilalta 2006}, ^{Ramirez-Marcial et al. 2010}, personal observation, RWJ). Study sites reported for H. longipes by ^{Jones & Baro-Peruyero (2002)} in the municipality of Tenejapa have also been deforested along the Yoshib River. Deforestation due to agricultural activities has occurred in the Tapachula area of Chiapas, where ^{Bodegas et al. (1977)} reported the presence of H. rovirosae and H. latifolia. Subsequently, populations of H. rovirosae have not been found. Several individuals of H latifolia were found near the Bodegas collection sites (^{Jones 1994}) but only within small forested fragments on the edge of cacao plantations or close to dwellings. Also, populations of H. latifolia in the lowland watershed of the Mapastepec River (MEXU, ^{Jones 1994}) are also in highly disturbed sites due to increased human populations along the river, and corresponding deforestation and agricultural activities that have intensified seasonal flooding and consequent damage to the floodplain.

Evidence of synchronous flowering (C.1.3), is reported by ^{Fryxell (1988)}; who gives general flowering periods for species, and writes that for the genus, “flowering is definitely seasonal” (^{Fryxell 1979}). ^{Jones & Baro-Peruyero (2002)} followed the phenology of populations of H. montebellensis, H. longipes and H. mexicanus in Chiapas and reported that flowering periods of both sexes occurred over a period of less than three months for these species. ^{Stansly (1985)} reported similar flowering periods for H. nutricia in Tabasco. However, H. trilobata has been collected with flowers in all months of the year (^{Herbario CICY 2018}), and flowering is apparently not synchronous and for extended months in H. rovirosae, H. intergerrima and H. tomentosa.

No data is available for the following genetic evaluation (Criteria C.2) for the amount of heterocigosity, genetic structure, amount genetic variation, and population differentiation; thus all criteria were assigned zero (Criteria C.2.1-C.2.4.). Five of the seven specialized biotic interactions (Criteria C.3) have not been reported for Hampea. There is no data or the response was negative for important interactions with nurse plants, a parasitic interaction with a host plant, a specific pollinator, obligate myrmecophilia, or strict dependence on mycorrhizal association. All species, except H. breedlovei, were observed in the field and there was no evidence of clonal reproduction.

The two specialized biotic interactions pertinent to species of Hampea have to do with specific dispersal agents and high biotic mortality agent for one species. Hampea is unique within the cotton tribe for the presentation of the seeds within a white aril in a hanging open, fruit capsule (Figure 2), a trait considered an adaptation to dispersal by vertebrates (^{Howe & Smallwood 1982}). Birds have been reported feeding on seeds of Hampea (^{Fryxell 1979}, ^{Marini & Cavalcanti 1998}) indicating that the dispersal of seeds apparently requires a frugivorous vertebrate (C.3.3 = 1). For another specific biotic interaction (C.3.7 = 1), ^{Jones & Baro-Peruyero (2002)} report that the herbivorous beetle, Anthonomus palmeri Jones & Burke (Curculionidae) can destroy over 90 % of the flower buds of H. longipes which may reduce plant fecundity.

Criteria D. Impact of Human Activity. Three lowland species are apparently favored by anthropogenic disturbance (Criteria D.1 = -1): H. nutricia, H. trilobata, and H. rovirosae (^{Fryxell 1969}, ¹⁹⁸⁸, personal observations, RWJ). In the case of H. nutricia, individuals are common on roadsides and often used as shade trees in patio and used in fence rows (^{Fryxell 1979}). Hampea trilobata and H. rovirosae are also favored by disturbance and occur on the edge of clearings near dwellings and roadsides. Although the lowland species, H. stipitata, is common along roadsides in the Lacandón region of northeastern Chiapas, it is apparently not benefited by disturbance (D.1 = 0), as are H. nutricia, H. trilobata, or H. rovirosae and is more restricted to less disturbed lowland and montane tropical rainforests (^{Fryxell 1988}). The final lowland species, H. latifolia is apparently negatively affected by disturbance (D.1 = 1), as it is confined to small forest remnants and riparian vegetation within the highly developed agricultural region of the Soconusco (^{Fryxell 1988}, ^{Jones 1994}).

Based on herbarium data and observations in the field, all remaining species were considered negatively affected by disturbance (D.1 = 1). This is especially evident in the montane species, H. breedlovei, H. montebellensis, H. intergerrima and H. longipes, which have not been reported from disturbed habitats. Given their rarity and their observed association with the relatively undisturbed habitats, the mid-altitude and dry forest species, Hampea mexicana and H. tomentosa are also considered negatively affected by human disturbance.

Anthropogenic disturbance of habitats (Criteria D.2) was considered to negatively affect to some degree all species of Hampea in Mexico, except H. trilobata. Of the most affected, Hampea breedlovei is restricted to cloud forest habitats of Northern Chiapas, of which few remain (^{Ramı́rez-Marcial et al. 2001}). ^{Cayuela et al. (2006)} estimated that there are only 4,650 ha of primary cloud forest left in northern Chiapas; and noted that all forests were outside of protected areas and subject to high levels of deforestation (D.2 = 3). Although Hampea longipes has a greater distribution than H. breedlovei, it is also present in the temperate and cloud forest habitats of northern Chiapas and thus, affected by high levels of deforestation (D.2 = 3).

Three species are found in areas of relatively high human population densities and heavily impacted by human activity (D.2 = 3), based on analysis by ^{González-Abraham et al. (2015)}. Hampea intergerrima is found in the heavily populated regions of the eastern portions of the Neovolcanic axis of central Mexico. Hampea mexicana is found in scattered forest and riparian fragments, principally within the central depression Chiapas that also has a high level of human impact and H. tomentosa, which is found on the relatively highly populated Pacific coast regions from Guerrero to southern Jalisco.

Hampea montebellensis is apparently restricted to the humid forest types within the vicinity of the Lagunas de Montebello National Park which have been significantly impacted (D.2 = 3). Although the park is a protected area, fires occurred in there during the El Niño years of 1998 and in 2003 (^{Román-Cuesta & Martínez-Vilalta 2006}). Deforestation has also occurred in order to facilitate tourist activities near lakes of the reserve, as well as from typical clearing of land for extraction and agricultural activities, although some reforestation efforts have been made (^{Ramírez-Marcial et al. 2010}, personal observation, RWJ).

Hampea latifolia is an endemic of the Soconusco region which is considered to have one of the highest levels of human impact within Mexico, principally associated with the tropical lowland forest (^{González-Abraham et al. 2015}). Because H. latifolia is a tropical forest species, the anthropogenic impact on this species is considered to be high (D.2 = 3).

The habitat where H. stipitata is found is primarily the Lacandón region of northeastern Chiapas. Although much of this region is now protected, deforestation from the 1960s to 1990s was estimated to be as high as 40 % (^{Bray et al. 2008}). Deforestation rates have dropped in the last 20 years but conservation initiatives are complicated by ethnic tensions, poverty and land-tenure problems (^{Calleros-Rodríguez 2017}). Populations of H. stipitata in central Chiapas (municipality of Berriozabal) are also threatened by high deforestation rates and population pressure in this region (D.2 = 3).

Although H. rovirosae and H. nutricia are tolerant of disturbance, their habitats are among the most impacted by human activity (^{González-Abraham et al. 2015}). These species are considered to be moderately affected by habitat loss in some populations (D.2 = 1). Hampea trilobata occurs in some of the least disturbed habitats of the Yucatán peninsula with low human impact on habitats (D.2 = 0).

We considered that all montane species, H. breedlovei, H. montebellensis, H. intergerrima and H. longipes, will be negatively impacted by the global effect on habitat deterioration in the future (Criteria D.3) due to climate change (D.3 = 1).). Climate models predict that of 68 % of Mexican cloud forests will be lost by 2080 (^{Ponce-Reyes et al. 2012}). The remaining species of Hampea were considered to be in habitats less affected by global effects or capable of adapting to changes given the seasonality of their present habitat (D.3 = 0).

The effect of direct use of species (Criteria D.4) on species of the genus Hampea is apparently minimal. The most cited direct use of most species is the fashioning of crude rope made from dried, unprocessed bark (^{Fryxell 1969}). This use is reflected in the common name of Hampea, “majagua,” which is often a general term for a variety of plants from which the bark is used for this same purpose (^{Fryxell 1969}, ^{Cobos et al. 2009}, Veláquez et al. 2009, ^{Aranguren & Márquez 2011}). The wood of Hampea trees is light and apparently not used much in construction (^{Fryxell 1969}). Thus, we considered there was no effect of direct use of species (D4), except for moderate use for rope of H. nutricia and H. trilobata (D.4 = 1). Also, we found no mention of the fruits being used for human consumption or evidence of species of Hampea being propagated ex situ or cultivated (D5 = 0).

Conclusions

Based on a detailed evaluation of the criteria of MER-Plantas (^{SEMARNAT 2010}), seven species of the genus Hampea in Mexico were in a category of risk. Three species were categorized as “Risk of Extinctión”: H. montebellensis, H. breedlovei and H. longipes. Of the other four species, Hampea latifolia was categorized as “Threatened” in our analysis and H. stipitata, H. mexicana, and H. tomentosa were categorized as “Subject to Special protection”. Formal requests will be made to SEMARNAT to change the status of these five species to their respective categories based on the results of our analysis. Also, the information presented in this study can be used to evaluate the species for inclusion in the IUCN Red List (IUCN 2017).

The first step in the conservation of H. breedlovei is to verify that the species still exists. The Cerro Ahk’ubal Nab in the municipality of Tila may be the only site left which may maintain populations of the species. Similarly, current population estimates are needed for the other at-risk species: H. montebellensis, H. longipes, H. latifolia, H. stipitata, H. mexicana and H. tomentosa. For all species, and especially those in the highlands of Chiapas, many of the populations are in areas presently under control of indigenous or independent political groups with which collaboration would be essential to initiate and establish in situ conservation efforts.

The three endangered species, H. montebellensis, H. breedlovei and H. longipes, could serve as flagship species for cloud forests habitats of Chiapas. This is especially pertinent for the appropriately named, H. montebellensis which is restricted to the Lagunas de Montebello National Park and surrounding areas. Information about the species could be provided to tourists and residents of the park through printed material or on-line, and individual trees could be identified in the field with permanent plaques highlighting the unique and endemic nature of the species. The species could also be used in the present reforestation programs for the park (^{Ramírez-Marcial et al. 2010}), given that other species of Hampea have shown potential in reforestation programs in Costa Rica (^{Marquis & Clark 1989}).

These five species need to be included in regional and national in situ conservation programs for the preservation of sites with populations of these species, as well as, ex situ conservation programs for the cultivation in nurseries and botanical gardens and the inclusion in germplasm preservation, and maintenance. However, little information is available for ex situ conservation programs for propagation of Hampea. The first author has grown several trees of H. trilobata from rootstocks and the Central American species, Hampea appendiculata (D.Sm.) Standley has been grown from seedlings in outdoor experimental plots (^{Marquis & Clark 1989}). ^{Fryxell (1979)} reports that, “based on limited data,” seeds of Hampea lose their viability in approximately a year, suggesting that they are intermediate between orthodox and recalcitrant seeds. Only two species were found to be in international seed banks, H. nutricia and H. appendiculata, in the Kew Gardens seed collection (^{Royal Botanic Gardens Kew 2018}). Much more information is needed for the successful inclusion species of Hampea in ex situ conservation programs and germplasm preservation.