Materials and methods

Study area. Our research took place along the western border of the State of Michoacan, in central Mexico. The study area lies within 20º 23’ 37” - 17º 53’ 50” N and 100º 03’ 32” - 103º 44’ 49” W, geographic coordinates. The whole surveyed area covered 39,315 km², equivalent to 67 % of the whole surface of the State of Michoacán (a bit larger than the surface of Quintana Roo State). This area represents the contact point of four large physiographic-geological units, namely: El Bajio, La Meseta Purepecha, El Valle de Tepalcatepec and Sierra Madre del Sur (^{INEGI 2014}). Our surveyed area is limited by the border of Jalisco and Colima; to the north it is limited by Guanajuato and Queretaro, to the southeast it borders with Guerrero and to the south with the Pacific Ocean. The Territory concerned, covers the main extensions where the analyzed ecosystem is present, located in the northeastern, western, central and southern portions of the state, as discontinuous bands interrupted by the mountain ranges of the Transmexican Volcanic Belt and the Sierra Madre del Sur (Figure 1).

The complexity of the study area provides large altitudinal variability (0 to 3,840 m); its main elevations, associated with volcanic summit, are: The Pico Tancítaro (3,840 m above sea level) and several summits of more than 3,000 m that occur throughout this system. In the Sierra Madre del Sur, the most notorious elevations are around 3,000 m asl. A half-slope of these systems, at about 2,400 m asl, start the ravines that continue in slopes that run towards the foothills and lowlands, being the ones that descend towards the valley of the Tepalcatepec as the most notorious (^{Garduño 2005}, ^{Mendoza et al. 2009}). All these hills, except those of the lowlands, are dominated by steep middle-gradient slopes, responsible for the marked vegetation contrasts that occur in the altitudinal transect lines.

Considering the classification of ^{Köppen (1948)} modified by ^{García et al. (1961}, ¹⁹⁷⁰⁾, the climatic subgroups present in the study area (^{Antaramián 2005}) and their distribution are: Aw: warm sub-humid (tropical rainy, with predominant rains in summer) in most of the Sierra Madre del Sur, in the coastal plain and in the southern slope of the Transmexican Volcanic Belt. BS: dry and semi-dry (dry steppe) in the depression of the rivers Tepalcatepec and Balsas. Cw: sub humid temperate (temperate with rain in summer) in the volcanic belt and in the associated plateaus. Cf: (temperate with rains throughout the year) in the highest parts of the volcanic summit.

Altitudinal variations and thereof climatic conditions, impact distribution of the vegetation. ^{Gopar-Merino et al. (2015)} conducted a bioclimatic mapping of the state by establishing the isobioclimates represented (sensu ^{Rivas-Martinez et al. 1999}). In general, these are tropical bioclimates with seasonal precipitation and Tropical Xeric, while at the Isobioclimates level there are 13 types present, which gives an idea of the great variability of vegetation and soils which occur along the altitudinal gradient.

Vegetation Analysis. Sampling units, hereafter denoted as relevés, were placed in representative rather homogeneous and well-conserved vegetation areas. For the conducting vegetation surveys and their subsequent analysis and interpretation, the postulates of the phytosociological method were followed (^{Westhoff & van der Maarel 1978}, ^{Braun-Blanquet 1979}). The phytosociological approach aims at providing rigorous vegetation description of plant communities and their distribution along environmental gradients (^{Rivas-Martínez et al. 1999}). Plant communities are eventually typified following protocols which must be accomplished to select diagnostic species and nomenclatural guidelines valid worldwide (^{Velázquez et al. 2016}). Methodologically, two phases must be implemented in phytosociological studies: firstly, data gathering in the field by means of surveys or relevés which consider complete species inventories and structural and ecological descriptions of each site; secondly data analyses in order to classify relevés and species into phytosociological packages by means of multivariate analyses. One of the most widely used statistical multivariate analytical packages is TWISNSPAN (^{Hill 1979}, now available in PC ORD Version 5). At these stage, generic plant communities turn into plant associations with a unique combination of diagnostic species and a unique name. Finally, it should be noted that the taxonomic and nomenclature aspects considered in the proposals are based on definitions, rules and recommendations of the International Code of Phytosociological Nomenclature (CNF) (^{Weber et al. 2000}). A thorough step-by-step explanation of the phytosociological approach may be consulted at ^{Velázquez et al. (2016)}. The concept of potential vegetation was also considered, (^{Tüxen 1956}, ^{Rzedowski 1978}, ^{Géhu 2006}), to dissociate close physiognomic formations of the Seasonally Dry Tropical Forest but with different ecological states. Finally, to add that for circumscribe the distribution of the associations have been considered the physiographic provinces and sub-provinces defined for the state by ^{INEGI (2014)}. Extensive recognition is first performed so that vegetation type borders are elucidated. Relevés in tree dominated areas covered 20 by 20 meters as suggested by ^{Kent & Coker (1992)}. Because of the floristic complexity and rich species composition, close by relevés were often surveyed to complete a robust description of the plant communities. A detailed floristic, phenological and physiognomic characterization is then performed at each relevé. Furthermore, ecological, geographical and emplacement data are also gathered. Structurally, the tree layer (above 5 m), the shrub layer (one to three m) and the herb layer (less than one meter) were first surveyed.

Due to the variety of forest types, we adopted the typology of ^{Rivas-Martínez et al. (1999)} namely: megaforest (taller than 50 m), macroforest (between 22 and less than 50 m tall), mesoforest (between 12 and less than 22 m), and microforest (between four to 12 m).

Each plant species was listed, and its total cover was estimated given their cover values in classes as suggested by ^{Braun-Blanquet (1979)} and adapted by ^{Van der Maarel (1979)}. The classes were: Class 9: 75 a 100 %; Class 8: 50 a 75 ; Class 7: 25 a 50 %; Class 6: 5 a 25 %; Class 1: abundant but cover of less than 5 %; Class +: occasional but cover of less than 5 %; and Class r: rare species.

Likewise, the diagnostic species were identified, by means of a synthetic analysis between the different groupings of inventories, which allowed pondering their frequency of appearance in each one of the communities as well as its ecological and distributional suitability.

Floristic analyses. In the field, plant species were identified to species level. This is done with the aid of field guides and expertise knowledge of the authors. In many cases, plant specimens were not identified, so that plant specimens were collected and brought to the herbariums for further identification with the aid of floristic experts. In some cases, the quality of the samples for identification was not possible so that rather than species name a genus of the plant specimen was assigned. The following regional floras were consulted in this process: Flora of the Valley of Mexico (^{Rzedowski 2001}), Flora of the Bajío and of adjacent regions (^{Rzedowski & Calderón de Rzedowski 1985-2015}), Flora Novogaliciana (^{McVaugh 1987}) and Tropical Trees of Mexico (^{Pennington & Sarukhán 2005}). The updates to the Mesoamerican Flora were also considered (www.tropicos.org/Project/FM). The group of Pteridophyta was determined following ^{Mickel & Beitel (1988)} and the families published in the Flora of the Bajio (Rzedowski & Calderón de Rzedowski op. cit.). The scientific names of the species follow the proposal of International Plant Name Index (www.ipni.org). Of the majority of the registered plants, a duplicate has been housed in the Herbarium of the Institute of Ecology, AC. (IEB-Pátzcuaro).

The field work was carried out in successive botanical campaigns developed between 1994 and 2013. In obtaining the floristic information we are aware of the possible absence of species of difficult detection as certain epiphytes or plants with optimal phenological in a period of time different to that in which the relevé was carried out; this should not be an obstacle to question the classification results obtained.

Information processing. The inventories considered were up-loaded into a gross table (Excel) for analysis and classification. Excel document further served as reference for conducting a multivariate exploratory analysis that allowed identifying the statistical weight that differentiates each group. This analysis was carried out with the PC-ORD package (^{Mccune & Mefford 1999}); in specific, the analyses used were principal components and canonical discrimination (^{Ludwig & Reynolds 1988}). The indexes obtained in the field were transformed to the scale proposed by ^{Van der Maarel (1979)}, and thus be able to submit it to the statistical analysis using the analysis PC-Ord of the Program TWINSPAN Version 5 (^{Hill 1979}). This method classifies species and samples, producing successive and hierarchical groups as levels of analysis advance; likewise, the process allowed generating dendrograms of the established groupings, which relate to the recognized communities. After the final arrangement of the inventories, the indexes were re-transformed to the Braun-Blanquet scale; thus, obtained tables served as reference for the diagnosis of the recognized associations. Furthermore, diagnostic species (characteristics and exclusive) were identified, by means of a synthetic analysis between the different groupings of inventories, which allowed pondering their frequency of appearance in each one of the communities as well as its ecological and distributional suitability.

Results

As a result of the field work in which 82 inventories were made, 1,491 plant specimens were collected and identified for compiling a list of 338 species and 131 genera belonging to 34 families. The vernacular name given by 47 collectors is also given. The information obtained from the 82 inventories served as a reference for the separation of the groups considered potential plant communities. Successive classification analyses allowed to separate the 9 groups of inventories that support the respective plant groups identified (8 associations and 1 community) (Figure 2). The results showed that communities follow a discontinuity gradient throughout the entire study area. In the analysis process, successive cutting levels (between 3 and 6) were carried out, which led to the identification of the relevés grouped per plant association/community.

From the groupings obtained, we analyzed the fidelity of the most representative species of each group by means of a comparison among the most frequent species. A first filter allowed selecting those that were dominant or exclusive on each situation, obtaining thus a filtered and ordered fidelity table (Table 1); it identifies the groups of species selected as diagnoses, which were considered later in the selection of characteristics and differentials of associations. The values of the crosses obtained between species and groups are differentiated in two terms: the first is the weighted frequency of the number of inventories where the species is present, while the second indicates the average cover of the species.

Based on the identified clusters and trends in the emergence of the diagnostic species and their indicator value, the tables of the recognized communities were organized. In most of them the reference information was considered enough to propose them as associations. For each situation the description has been structured in a series of attributes related to physiognomy, structure and floristic composition, ecology and distribution, variability and observations. Next to the name of the proposed association/community, the number of the reference phytosociological table and the number of the selected type inventory are attached. The tables were continuously placed after the diagnosis of the associations. Full description of the nine plant associations/communities is provided in following paragraphs.

Lysilomo acapulcensis - Heliocarpetum terebinthinacei (Table 2, inventory n^o 3 intable 2). Physiognomy, structure and floristic composition.- Deciduous Micro-forest between 6 and 12 (15) m of height, dominated by trees microphyllous, and, to a lesser extent, broad-leaf and thorny, profusely branched, sun shielded and twisted plant species with frequent epiphytes and lianas. It has an average cover of 80 %, with a shrub layer of heterogeneous cover that sometimes extends in an intermediate (sub arboreal), which hinders its differentiation. The herbaceous stratum covers much of the understory with cover around 80 % and can be dominated by species indicating a certain disturbance. As characteristic species of the association the following have been selected: the Heliocarpus terebinthinaceus Hochr, Lysiloma acapulcense (Kunth) Benth., Ipomoea murucoides Roem. & Schult., Vachellia pennatula (Schltdl. & Cham.) Seigler & Ebinger and Eysenhardtia polystachya (Ortega) Sarg., which together with Bursera spp., Ceiba aesculifolia Britt. & Baker f., Zanthoxylum fagara Sarg. and Forestiera phillyreoides Torr. dominate the arboreal canopy. Occasionally, those species may be present also in the understory, where Croton sp., Opuntia sp., Vachellia farnesiana (L.) Wight & Arn, Tecoma stans (L.) Juss. ex Kunth, Celtis caudata Planch. and Mimosa sp. are the dominant species. The herbaceous stratum is very heterogeneous and diverse in terms of its composition and participation, without a clear group of recurring species; in any case, representatives of the genera Salvia L., Senecio L., Aegopogon, Lasiacis (Griseb.) Hitchc. and Heimia salicifolia Link stand out.

Ecology and distribution.- The characteristic species of this association have an endemic distribution to Mesoamerica, with the exception of the representatives of Heliocarpus L. and Eysenhardtia Kunth, which are exclusive to Mexico. Its presence is linked to hills and slopes of the physiographic sub-provinces Sierra and Bajio Michoacano and Chapala, to the north and northwest of the state; the geographical exception includes an inventory located in the southern coastal Mountain Range sub-province (Table 2, Inventory 4), which was included in this block of inventories during the grouping analysis.

The altitudinal range fluctuates between the dimensions of 1,300 and 2,200 m; towards the latter its presence is restricted to sunny spots and rocky inclined slopes. This association distributes within the Dry superior Thermotropical, between the Mesotropical and Low sub-humid bioclimatic belts.

Variability and observations.- Inventories 1 and 10 represent situations marked by the notable discontinuity of the arboreal stratum due to tree extraction. On the other hand, inventories 9 to 11 have a low representation of characteristic species, which did not prevent the statistical analysis from linking them with other inventories. In the area of potential distribution of the association, linked to abandoned agricultural land or cleared forests, there is a high thorny thicket of secondary trait dominated by representatives of Vachellia Wight & Arn., Mimosa L. and Opuntia Mill., as well as Grasses (Aegopogon Humb. & Bonpl. ex Willd., Lasiacis (Griseb.) Hitchc.) and used extensively by cattle. The unit of thorny forest defined by ^{Rzedowski (1978)} for the zone would be included in this association, as a stable and permanent variant of secondary character.

Ceibo aesculifoliae-Lysilometum divaricatae (Table 3; Inventory n^o 3 intable 3). Physiognomy, structure and floristic composition.- Deciduous Micro-forest, between 7 and 15 m high, with a physiognomy similar to that of the previous association, with occasional nubs protruding from the canopy. Ceiba aesculifolia Britt. & Baker f., Lysiloma divaricatum (Jacq.) Benth., Ipomoea murucoides Roem. & Schult. and Zanthoxylum fagara Sarg, dominates its arboreal stratum; the first three are proposed as characteristics of the association, together with Pseudobombax ellipticum (Kunth) Dugand. In this stratum can also be present Bursera spp., Vachellia pennatula (Schltdl. & Cham.) Seigler & Ebinger, Heliocarpus terebinthinaceus Hochr., Cardiospermum halicacabum L., Celtis caudata Planch, Eysenhardtia sp., Ipomoea sp., Stenocereus queretaroensis (F.A.C. Weber) Buxb., as the most representative species. In the shrub stratum, the best-represented genera are Opuntia Mill., Vachellia Wight & Arn., Agave L., Randia L., Montanoa Cerv., Salvia L. and Ageratina Spach with an average cover of 50 %; as in the previous case, elements of the arboreal stratum can be present. The herbaceous stratum, of very heterogeneous composition and with an average cover of 60 %, is dominated by a few species in most of the different cases, although with a predominance of the compound group.

Ecology and distribution.- The association is distributed mainly at the northwest of the state, linked to slopes and hills of low or medium gradient, on low developed soils. From the physiographic point of view, it is present in favorable enclaves of the sub-provinces Chapala, Sierra and Bajio Michoacano, Coastal Mountain Range of the South and, punctually, Trans-Mexican Volcanic Belt and Southern Border Escarpment. The altitudinal range of distribution in these units oscillates between 900 and 1,900 m. From the bioclimatic point of view, these are the dry thermotropical soils and to a lesser extent the lower Mesotropical horizon of dry ombrotype, where most of the backup inventories are located; punctually it can arise also in favorable enclaves of the Lower Sub-humid Lower Mesotropical horizon.

Variability and observations.- The community is affected by the same processes of alteration commented for the previous association which lead to the presence of secondary shrubs for cattle use, which form mosaics with it.

The three communities described below come from the information published by several of us (^{Pérez-Vega et al. 2010}) for the San José de Chila river basin (Sierra Madre del Sur, Michoacán), which addresses the attributes that best explain the patterns that differentiate the plant communities found (lithology, altitude and soil depth). The information in the table was transcribed into the general reference table; indicate that this information lacks data relating to herbaceous and subshrub strata (< 3 m). From the physiographic point of view, these sites are located within the Southern Coastal Mountain Range Sub province. The main differences between these communities are related to their structure, floristic composition, and altitudinal range of distribution and bioclimatic affiliation. Next, a characterization of the four referred communities is presented:

Caesalpinio platylobae - Cordietum elaeagnoidis (Table 4; Inventory type n^o 5 intable 4). Physiognomy, structure and floristic composition.- Deciduous Micro-forest, between 6 and 10 m high, formed by leafed trees and deciduous winter microphyllous, several of them thorny. Its arboreal stratum has an almost continuous canopy dominated by Cordia elaeagnoides A. DC., C. dodecandra DC., Caesalpinia platyloba S. Watson, C. eriostachys Benth., Apoplanesia paniculata C. Presl and Bursera coyucensis Bullock, species selected as association characteristics. Other arboreal species of more sporadic presence are: Tabebuia ochracea (Cham.) Standl., Malpighia mexicana A. Juss., Pachycereus pecten-aboriginum Britton & Rose and Caesalpinia velutina Standl. In the shrub stratum, Erythroxylum mexicanum Kunth, E. rotundifolium Lunan, Croton flavescens Greenm., and Manihot intermedia Weath. are usually present.

Ecology and distribution.- Of the four communities belonging to the basin, this is the one at a lowest altitude, between 250 and 500 m, being this dominant factor against others such as the lithological or the depth of the soil. Most of the inventories are oriented to the west and are on slopes of steep gradients, between 20 and 40 %. The bioclimatic floor (Horizon) where it is located corresponds to the lower dry upper infratropical.

Cochlospermo vitifolii - Lueheetum candidae (Table 5, inventory type n^o 5 intable 5). Physiognomy, structure and floristic composition.- Deciduous Micro-forests with an average height of 14 m, although they can reach up to 20 m (Meso-forest). A tree stratum dominates it with a cover close to 100 %, where abundant leafed elements with some thorny and microphyllous can be found, among which are Luehea candida Mart., Cochlospermum vitifolium (Willd.) Spreng., Lysiloma divaricatum (Jacq.) Benth., Poeppigia procera C. Presl and Tabebuia ochracea (Cham.) Standl., all of which were selected as characteristics of the association. Other elements that may appear in the community are Eysenhardtia polystachya (Ortega) Sarg., Lonchocarpus caudatus Pittier, Bursera grandifolia (Schltdl.) Engl. and Cordia elaeagnoides A. DC.

Ecology and distribution.- It is distributed largely above the previous community, between 330 and 800 m of altitude, on moderate slopes (between 20 and 40 %) and oriented preferably to the South. This site corresponds to the lower dry upper infratropical soil (Horizon), although inventories made at higher elevations can be linked to the Thermotropical soil the lower horizon.

Variability and observations.- In the original classification analysis of ^{Pérez-Vega et al. (2010)} two sub communities associated with the different lithology were differentiated. One with Ceiba aesculifolia Britt. & Baker f. and Lonchocarpus caudatus Pittier corresponding to soils developed on granite; the other with Poeppigia procera C. Presl and Eysenhardtia polystachya (Ortega) Sarg., on schists. However, in the general context of this study, such differentiation was not recognized due to the low weight of the inventories and the species used for it.

Lysilomo divaricatae - Cordietum elaeagnoidis (Table 6, Inventory type n^o 2 intable 6). Physiognomy, structure and floristic composition.- Deciduous Micro-forest between 8 and 16 m of height, dominated by leafed elements and microphyllous, and with the presence of some species with exfoliating stems. It has a tree stratum with cover around 80% that is dominated by Cordia elaeagnoides A. DC., Lysiloma divaricatum (Jacq.) Benth., Fridericia viscida (Donn. Sm.) L.G. Lohmann and Heliocarpus donnellsmithii Rose, which have been selected as characteristics of the association. Trees and tall shrubs, some spinous, such as Bursera spp., Randia spp., Lonchocarpus hintonii Sandwith, Haematoxylum brasiletto H. Karst., Luehea candida Mart, Tabebuia ochracea (Cham.) Standl., Cochlospermum vitifolium (Willd.) Spreng. and Apoplanesia paniculata C. Presl, may also occur sporadically.

Ecology and distribution.- It occupies a strip between the altitudes of 350 and 1,000 m on slopes moderately inclined (between 20 and 30 %) oriented preferably to the south. Its physiognomic characteristics, its bioclimatic site and its framing in the vegetation units considered match those of the previous association.

Comunity of Spondias purpurea-Cochlospermum vitifolium (Table 7). Physiognomy, structure and floristic composition.- Deciduous Micro-forest, poor in species, with an average altitude of 13 m of height and with a cover around 80 %, dominated by leafed or without awn microphanerophyte. In addition to the species that give the name to the community, other species of the arboreal and shrub strata may be present, such as: Ceiba aesculifolia Britt. & Baker f., Cordia elaeagnoides A. DC., Erythroxylum mexicanum Kunth, Ipomoea arborescens Sweet, Fridericia viscida (Donn. Sm.) L.G. Lohmann, Amphipterygium adstringens (Schltdl.) Standl., Bursera instabilis McVaugh & Rzed., Lonchocarpus eriocarinalis Micheli and Randia spp.

Ecology and distribution.- The community has been registered between the altitudes of 410 and 710 m, with a majority exposure to the south and on slopes with an average gradient of 30°. Its physiographic and bioclimatic positions remain similar to the previous association described.

Stenocereo quevedonis - Cordietum selerianae (Table 8, Inventory type n^o 3 intable 8). Physiognomy, structure and floristic composition.- High shrub dominated by deciduous and spiny elements with height ranging from 4 to 10 m, which leads to classifying it. The upper stratum has a low and discontinuous coverage, around 40 % and highlights the columnar eminences of the Pitaya cactus (Stenocereus quevedonis (J.G. Ortega) Buxb.); This heterogeneity favors the entry of light into the shrub stratum that reaches an average cover close to 70 %; Usually the two strata are interspersed making their differentiation difficult, and their species can participate interchangeably in both. The more conspicuous arboreal elements are: Cordia seleriana Fernald, Stenocereus quevedonis (J.G. Ortega) Buxb., Apoplanesia paniculata C. Presl, Randia capitate DC., Amphipterygium glaucum (Hemsl. & Rose) Hemsl. & Rose, which together with the shrub Salpianthus arenarius Bonpl., were selected as characteristic of the association. Other low trees or shrubs with sporadic presence in the community are Lonchocarpus eriocarinalis Micheli, Amphipterygium adstringens (Schltdl.) Standl., Gliricidia sepium (Jacq.) Kunth ex Walp., Haematoxylum brasiletto H. Karst., Lysiloma divaricatum (Jacq.) Benth., Cordia elaeagnoides A. DC., Erythroxylum rotundifolium Lunan, Bursera trimera Bullock, Bursera sarukhanii Guevara & Rzed. and Cordia nelsonii I.M. Johnst. In the shrub stratum the most constant species are Salpianthus arenarius Bonpl., Eysenhardtia sp., and Opuntia sp.; others of more sporadic appearance are Croton sp., Randia spp., Bursera spp., Fouquieria formosa Kunth and Colubrina heteroneura Standl. In the herbaceous stratum, elements of the families Acanthaceae and Compositae prevailed.

Ecology and distribution.- The association is presented in the lower and arid regions of the Tepalcatepec River basin, between 150-400 m altitude, within the physiographic subregions of the Tepalcatepec depression and the Coastal Mountain Range of South. It occupies stony slopes with moderate gradient oriented preferably to the N. From a bioclimatic point of view, it has its optimal distribution in the upper Infratropical horizon, under the upper and lower dry semiarid ombrotypes.

Guazumo ulmifoliae - Cordietum elaeagnoidis (Table 9; Inventory n^o 2 intable 9). Physiognomy, structure and floristic composition.- Low forest or high shrub dominated by deciduous microfilated and spiny free species, which form a semi-continuous canopy with a height between 5 and 15 m, which allows it to be placed in the Micro-forest category. In comparison with other neighboring communities, this presents, in most cases, a notable poverty of species, being Cordia elaeagnoides A. DC., Lysiloma divaricatum (Jacq.) Benth., Guazuma ulmifolia Lam. and Crescentia alata Kunth the most recurrent, selected as characteristics of the association, while being at the same time the dominant ones of the arboreal stratum, in particular the first two. In the shrub stratum of these communities, spinous and crasicaules species are sporadically present, such as Opuntia spp., Stenocereus quevedonis (J.G. Ortega) Buxb., Acanthocereus sp., Cardiospermum halicacabum L., Bursera spp., Backebergia militaris (Audot) Bravo ex Sánchez-Mej., Acaciella angustissima (Mill.) Britton & Rose, Senna atomaria (L.) H.S. Irwin & Barneby, Heliocarpus sp., Caesalpinia platyloba S. Watson, Alvaradoa amorphoides Liebm., etc. In the herbaceous stratum elements of the families Compositae and Acanthaceae are present.

Ecology and distribution.- Samples of the community have been taken between 180 and 620 m of altitude, on stony slopes of moderate gradient; it is distributed preferably in the subprovince physiographic of the Tepalcatepec depression and in a marginal form in the coastal mountain range of the south.

It is recognized in the inventories 5 and 6, a group of exclusive species that could denote some ecological factor, and that are Bursera palmeri S. Watson, Backebergia militaris (Audot) Bravo ex Sánchez-Mej. and Acaciella angustissima (Mill.) Britton & Rose var. angustissima.

Lonchocarpo huetamoensis - Cordietum elaeagnoidis (Table 10; Inventory^o 4 in table 10). Physiognomy, structure and floristic composition.- Low deciduous forest with a dense and multi-specific arboreal stratum between 10 and 15 m of height, which allows it to be placed in the Micro-forest category (Meso-forest). The dominant species are: Stenocereus quevedonis (J.G. Ortega) Buxb., Lonchocarpus huetamoensis M. Sousa & J.C. Soto, Tabebuia impetiginosa (Mart. ex DC.) Standl., Cyrtocarpa procera Kunth, Lysiloma tergeminum Benth., Coccoloba acapulcensis Standl., all of them selected as characteristics of association; other species present in this stratum are Lysiloma divaricatum (Jacq.) J.F. Macbr., Ceiba aesculifolia Britt. & Baker f., Erythrina sp., etc. In the shrub or subarboreal stratum, the previous species and others may participate, such as: Randia spp., Erythroxylum rotundifolium Lunan, Euphorbia tanquahuete Sessé & Moc.and other representatives of the genera, mainly: Croton L., Albizia Durazz., Triumfetta L. and Lantana L. The herbaceous stratum presents variable cover of up to 50 % or more and is quite heterogeneous in its composition with predominance of Leptochloa P. Beauv., Elytraria Michx., Aegopogon Humb. & Bonpl. ex Willd. and several species of the family Acanthaceae, Malphigiaceae and Malvaceae.

Ecology and distribution.- The association is described as belonging to middle and low slopes with moderate gradients, found at the municipality of La Huacana, in depression of the Balsas River. From the physiographic point of view this location belongs to the subprovince Coastal Mountain Range of the South. The altitude in which the strip is present oscillates between 330 and 520 m and the most frequent exposure is N. The bioclimatic floor which best fits the association is the upper Infratropical of Dry ombrotype, being able to present also in the upper Semiarid upper Infratropical, and in the low Dry Thermotropical.

Variability.- Two restricted species packages are recognized respectively in the inventories 1, 2 and 3, of the first, and 4, 5 and 6 of the second. From the first, the most frequent exclusive species are the “Pachón", an unidentified arborescent cactus, Vitex mollis Kunth, Karwinskia humboldtiana (Schult.) Zucc., Randia sp., Cordia morelosana Standl. and Coursetia sp. From the second group Apoplanesia paniculata C. Presl, Heliocarpus occidentalis Rose, Bursera trimera Bullock and B. fagaroides Engl. are typical. Both variants allow intuiting a differentiated behavior motivated by some unkwon factor of the environment.

As a biogeographical affiliation and considering the associations in the vegetation units proposed by ^{Rzedowski (1978)}, and its distribution within the physiographic sub provinces of ^{INEGI (2014)}, the following correspondences are recognized: Tropical Deciduous Forest of Bajio: Lysilomo acapulcensis-heliocarpetum terebinthinacei and Ceibo aesculifoliae-Lysilometum divaricatae. Tropical Deciduous Forest of the Tepalcatepec Valley: Lysilomo divaricate-Cordietum elaeagnoidis. Tropical Deciduous Forest of the Coastal Mountain Range of the South: Ceibo aesculifoliae-Lysilometum divaricatum, Caesalpinio platylobae-Cordietum elaeagnoidis, Cochlospermo vitifolii-Luheetum candidae, Lonchocarpo huetamoensis-Cordietum elaeagnoidis and Community of Spondias purpurea L.-Cochlospermum vitifolium (Willd.) Spreng.. Xeric Thicket of the Valley of Tehuantepec: Stenocereo quevedonis-Cordietum selerianae.

The nine Phytosociological tables supporting the associations and communities described, resulting in the process of sorting and classification as shown in Figure 2, are presented next.

Sintaxonomic Scheme. The proposed taxonomic units are included in the provisional phytosociological class of Pachycereo pecten-aborigini-Lysilometea divaricati (^{Peinado et al. 2008}), which gathers tropical deciduous forests and thorny forests and shrubs, distributed in the Xerophytic-Mexican and Caribbean region (sensu ^{Rzedowski 1978}). Bioclimatically they cover the lower Semiarid ombrotypes (Spiny shrubs), upper Semiarid (Spiny Forests) and Dry (Deciduous Forests); the thermotypes covered are Infratropical and Thermotropical (^{Macías-Rodríguez et al. 2014}).

In the absence of more ph ytosociological information of the surrounding territories to the one considered here, the correspondence of the associations described, and its floristic and ecological features, along with the units of higher range with which we propose their following phytosociological classification and phytocenotic diversity array (Table 11).

Discussion

The present research allowed to characterize and to compare attributes such as phenotype, structure, floristic composition, ecological affinities and distribution to define (dis)similarities among association/community. Our outcomes unravel two orders, three alliances and eight associations. The novel phytosociological classification here presented and its related syntaxonomical scheme (Table 11) comprised the choice of the characteristic and exclusive species of each association, its preferred environments and its physiographic and bioclimatic jurisdiction. In addition, one community (Spondias purpurea and Cochlospermum vitifolium) was described and further detailed floristic surveys must be conducted to eventually name this as a new association.

The first two years of intensive geobotanical survey were not enough to distinguished phenotypic and floristic contrasts among associations but orders. Alliances were further distinguished in the next five years by revisiting areas as well as adding sampling units and deepen on floristic inventories. A few additional years of sampling and long-term phenotypic expressions derived from outstanding dry and wet climatic seasons were crucial to eventually confirm the phytosociological scheme here presented at association level. This is clearly the case of the Lysilomo acapulcenis-Heliocarpetum terebinthinacei and Ceibo aesculifoliae-Lysilometum microphyllae associations; which phenotypes remain as rather tropical dry on temperate sub-humid or temperate dry climatic conditions accordingly to ^{García (2004)}. These two associations in wet climatic seasons truly change their physiognomy denoting adaptation attributes such as reducing substantially the proportion of leaves to be dropped in heavily rainy seasons of consecutive years.

Another long-term observation and vegetation affinity concern the Spondias purpurea and Cochlospermum vitifolium community. Floristically, this community was very consisting throughout the years. Geoedaphical conditions and degree of human disturbance play a role to denote a common physiognomic variation closely related to the vegetation type known as Xerophytic Shrubland.

We regard out outcomes as complete, phyotosociologically, for the Seasonally Dry Tropical Forest of Michoacan. Azonal associations limited by site or local outstanding conditions are yet to be described. There may be a couple more zonal associations yet to be described in neighboring areas with Guerreo and Colima states, especially along costal landscapes. ^{Hernández-Toro (2003)} contributes a series of inventories of the area of Cabo Corrientes (Jalisco) linked to the Seasonal Dry Tropical Forest where the dominant tree species present are Amphipterygium adstringens (Schltdl.) Standl., Plumeria rubra L., Caesalpinia eriostachys Benth. and Lysiloma divaricatum (Jacq.) Benth. to a lesser extent, among others, Piranhea mexicana (Standl.) Radcl.-Sm., Lonchocarpus luteomaculatus Pittier, Pachycereus pecten-aboriginum Britton & Rose, Bursera vazquezyanesii Rzed. & Calderón or Tabebuia rosea DC. Although it shares with our inventories some species and a similar structure, they also have in common the great diversity of species in the understory strata, which includes many species present in very few inventories, preventing the establishment of phytosociological similarities with the associations proposed here. It is likely that at the level of order or phytosociological alliance, links can be established. However, the collection of the field information carried out by ^{Hernández-Toro (2003)} with inventories by strata and without establishing continuity between them, considerably hinders the phytosociological interpretation of the data. As we have observed, certain changes in environmental conditions favor the transit of this formation towards Spiny Forest and Sub-deciduous Tropical Forest.

The present phytocenotic diversity comprised in the Seasonally Dry Tropical Forest (SDTF) of western Michoacán contrasts with traditional classification schemes (Table 12). Low-height caducifolious jungle (^{Miranda & Hernandez-X 1963}) and Tropical caducifolious forest (^{Rzedowski 1978}) are the most outstanding denominations of the SDTF plant formation. ^{González-Medrano (2003)} distinguished three subformations, namely: Low-height caducifolious forest, Tropical dry forest, High succulent spiny scrubland. ^{INEGI (2014)}, furthermore, in a crucial effort to construct a hierarchical vegetation scheme for the National Forest Inventory distinguished three subformations, namely: Subtropical Scrubland, Low height caducifolious jungle and Single steam succulent scrubland. These four previous classification schemes at national level follow untraceable attributes to define their vegetation classes, levels of aggregation and nomenclature principles. Deep floristic understanding as well as ecological characteristics are the common attributes considered by these authors. ^{Velázquez et al. (2016)} provided a framework to gain from these previous vegetation schemes and order physiognomic, climatic, phenotypic, botanical and floristic attributes to hierarchically organize and named global vegetation types. A synthetic correspondence of the present phytocenotic diversity comprised in the Seasonally Dry Tropical Forest (SDTF) of western Michoacán and the ones provided by these five previous authors was comprised in Table 12.

Regional studies to further split SDTF do not exist, nor in Jalisco, neither in Guerrero nor Oaxaca states where the same SDTF extends its distribution range. Authors working for decades in Chamela (^{Bullock et al.1995}, ^{Burgos & Maass 2004}), central Mexico (^{Trejo & Dirzo 2000}) and Puebla (^{Valiente-Banuet et al. 2000}) have deepen in the description of floristic aspects of the SDTF plant formation. These studies have focused in documenting the astonishing floristic diversity, rarity and endemicity of plant species harbored in the SDTF plant formation. ^{Valiente-Banuet et al. (2000)} provided detailed description of plant associations within the SDTF, namely “Cuajiotal”, “Izotal”, “Mezquital” and their variations within the Tehuacán-Cuicatlan study area. The hardly noticeable dominance of species and the smooth gradient of change in the SDTF has discourage most scholars to formally described distinguishable and discrete plant communities. All the previous authors have chiefly provided the floristic basis to attend the second phase of a sound vegetation study so that plant communities may be soon will be described for the entire distribution range pf the SDTF. In this sense, the present contribution may serve as a guideline to advance in the pending task concerning the complete inventory of plant communities contained in the Mexican SDTF.

At the local level, these communities present different dynamic stages, often interspersed, resulting from the unequal incidence of human activities, especially the extraction and those associated with grazing, leading to the establishment of a continuous decline in space (^{Burgos & Maass 2004}). This disturbance hinders the discrimination of these communities using images obtained by remote sensing. Several authors say that more than 80 % of the current area has been or is used for migratory cultivation and extensive livestock production (^{Trejo & Dirzo 2000}, ^{Sánchez-Azofeifa et al. 2005a}, ^2005b, ²⁰¹⁴). This incidence in the conservation state is reflected, in some of our inventories, in the dominance of indicator species, especially in the shrub stratum, in addition to the presence of obvious signs such as stumps, sprouts or cattle trails. The knowledge of the successional dynamics of these forests, and the incidence and degree of disturbance they present, is essential for the development of preservation strategies (^{Burgos & Maass 2004}, ^{Quesada et al. 2009}). Although the local inhabitants constitute the moving force behind such a situation, the underlying triggers and the controllers responsible for the current state of the SDTF have not been determined. ^{Castillo et al. (2005)} blame the model of agricultural invasion promoted by the Mexican government for decades, with the aim of seeking the expansion of profitable uses of the land. The alternative social explanation points to recurrent patterns of extreme poverty, inequality of opportunities, organized drug cartels, and the general lack of governability. In our opinion, these socioeconomic factors are probably the most important underlying forces that cause the disturbance of the SDTF.

In short, most Mexican SDTF are disturbed and their use for production and conservation are contested management alternatives. The productive use of the land is important to reduce poverty and inequity, while conservation is a priority since the SDTF represents a real genetic bank. However, the extraction of wood is technically absent since the few species of trees with commercial value (for example the genera Bursera, Cordia, Heliocarpus, Ipomoea, Lonchocarpus, Lysiloma, Tabebuia) are not profitable nor worth investing in.

Systematic studies of plant communities comprised in Seasonally Dry Tropical Forest is yet a pending task, globally. “Caatinga” and “Cerrado” in Brazil; “Bosque Bajo Andino” and “Bosques costeros del Caribe” in Colombia and other local nomenclature used for similar ecosystems in Africa and Asia are simple examples of the relevance of adopting a unifying framework to classify and name plant communities of this most threaten ecosystems. The present contribution was largely slow-down by the lack of complete floristic inventories and the biased geographic distribution of the floristic surveys. Semi-desertic conditions prevailed for most of the year and when rains fall, these are heavy, short, and catastrophic in most valleys. It was therefore that we needed over 20 years to make sure that our associations exist. We trust that further studies and especially public environmental policies to consider the outstanding phytocenotic diversity comprised in the Seasonally Dry Tropical Forest. This ecosystem looks homogenous, species- poor, resourceless, and unpleasant in the dry season but it is rather heterogenous, species-rich, resourceful and fascinating in the rainy season.