Introduction

Synanthropic vegetation thrives under human disturbance. Its composition is determined by climate and soil, but also by the type and intensity of disturbance (^{Rzedowski & Rzedowski, 2004}). The species of these communities - weeds - are sometimes undesirable and can cause considerable economic harm (^{Hamill et al., 2004}). However, they also have important ecological functions (^{Vibrans, 2016}).

Weeds are commonly defined as “plants growing where they are not wanted, or undesirable” (^{Monaco et al., 2002}); however, this definition is subjective, ambiguous, and not useful for understanding the biology of this group of plants. Here, we use the ecological concept: weeds are those species that can establish populations (obligatorily or facultatively) in places strongly disturbed by humans, without being cultivated. These may be cultivated fields (segetal or agrestal weeds) or ruderal environments (e.g., roadsides and field edges, surroundings of human settlements, urban vegetation, or vacant lots) (^{Baker, 1967}; ^{Rzedowski & Rzedowski, 2004}). This definition of a weed species helps delimitation, though the degree of human disturbance remains a source of imprecision.

Synanthropic plants have unusual dynamics, as they converge, diverge, and evolve with crops; they depend on disturbance to stabilize their communities (^{Mahaut et al., 2020}). However, many weeds inhabit not only environments disturbed by humans, but also by nature (e.g., water, landslides, tree falls, floods, hurricanes, fires, animals). Many species are even known from stable natural vegetation, although data on the proportions are lacking. Apparently, weedy species in natural vegetation are observed particularly in regions where weeds co-evolved with agriculture (^{Zohary, 1973}), though good documentation is lacking.

Separating weeds from non-weeds is difficult, given the overlaps in nature, but it is a necessary task for studying these species as a group. Generally, classification as a weed is based on observations and unquantified herbarium data. It tends to be arbitrary and non-reproducible. Thus, it is desirable to develop clear criteria for classifying a species as a weed. Then, the ecology, biogeography, evolution, and interaction with human activities of these species can be explored. A significant number of native weeds have evolved in Mexico and its cultural region, Mesoamerica, which includes parts of Central America. According to ^{Espinosa-García et al. (2004a}, ^b), 2,814 species are considered weeds, of which 2,197 (78%) are native.

Asteraceae Bercht. et J. Presl (Compositae Giseke nom. alt.) is representative of plant diversity in Mexico (^{Villaseñor et al., 2007}). It is the most species-rich family in the country with 3,113 species and endemism of 63.9% (^{Villaseñor, 2018}). Composites include a wide array of growth forms, which are distributed throughout the territory and all habitat types, including disturbed ones (^{Rzedowski, 1991}). Based on morphological and molecular characters, the family is organized into tribes (^{Villaseñor, 1993}); the most species-rich tribes in Mexico are Heliantheae, Eupatorieae, and Astereae (^{Villarreal & Villaseñor, 2004}).

The state of Michoacán ranks fifth in Asteraceae richness, with 837 species. This richness is due to variations in climate, soil, and topography, as well as its position on the limits of several biogeographic regions (^{Villaseñor, 2018}). Of these 837 species, one-third -282- have been reported previously as weeds (^{Villaseñor & Espinosa, 1998}), positioning Michoacán as the state with the highest number of reported weeds, followed by Jalisco.

Here, we provide an updated list of Asteraceae weeds in Michoacán. The selection was based on standardized criteria. We quantify the proportion of growth forms, habitats, and biogeographic aspects for weeds that are endemic to Mexico, native but not restricted to Mexico, and introduced, as well as the floristic similarity of the weeds of Michoacán with other states of Mexico. In addition, the proportion of weeds per tribe is considered.

Materials and methods

Michoacán lies in the central-western part of Mexico (20°23’38.40” - 17°54’54.00” N, 100° 3’46.80” - 103°44’16.80” W), bordering Jalisco and Guanajuato to the north, Querétaro in the northeast, Estado de México and Guerrero to the east, the Pacific Ocean to the south, and Colima and Jalisco to the west (Fig. 1). It has a surface area of 58,599 km², 3% of the country (^{INEGI, 2015}).

Figure 1 The state of Michoacán and its biogeographic provinces (modified from Conabio, 1997). 

The Nearctic and Neotropical biogeographic regions converge in the study area (^{Conabio, 1997}), as well as 4 biogeographic provinces, the Volcanic Belt, Pacific Coast, Balsas Depression, and the Sierra Madre del Sur (Fig. 1; ^{Conabio, 1997}). Also, it has a variable topography. As a result, it is one of the most phytodiverse states (^{Villaseñor & Ortiz, 2014}), and the 5 main biomes recognized for Mexico occur in Michoacán.

Mountainous terrain covers 62% of the territory. The highest elevation is the Pico Tancítaro (3,857 m); the coastline is 208 km long (^{Flores & Priego, 2011}). Average annual rainfall varies from 400 to 2,000 mm (^{Vidal-Zepeda, 1990}). A warm sub-humid climate (Aw) dominates in the central and southern parts (^{García, 1998}), with seasonally dry tropical forests occupying 50% of the state surface (^{Villaseñor & Ortiz, 2014}). Luvisols are the most important soil type, followed by vertisols and another 14 soil types (^{INEGI, 2014}, based on the World Reference Base for Soil Resources). Volcanic subsoil from the Cenozoic is found on 33% of the state’s surface; sedimentary and metamorphic rocks predominate in the south (^{Marín & Torres-Ruata, 1990}). Rainfed and irrigated agriculture occupies about 35% of the area; in the tropics, cattle pastures are common and cover 6.2% of the state’s area (^{INEGI, 2016}).

We used primary and secondary sources of information to identify weed species. Primary data sources were herbarium labels of field collections. Problems may arise in the acquisition and interpretation of label data since not all collectors include habitat information or only record the biome without indicating the degree of disturbance. The habitat information may be found in various label fields like observations, ecology, geoform, microenvironment, etc., which makes systematization more difficult (^{James et al., 2018}; ^{Moreno & Allkin, 1988}).

Secondary data sources were mainly floras and taxonomic treatments that included information on habitat. The origin of this information was usually not specified; it might also have been based on herbarium specimens or simply on observations by the specialists (Appendix).

Preliminary list. We worked with a database of all Asteraceae from Michoacán, managed and curated by one of the authors, an Asteraceae specialist (Villaseñor, unpublished data). It consisted of 13,641 records and 785 species (Fig. 2). The data derived mainly from the collections of the National Herbarium (MEXU) but included partial information from the herbaria CHAPA, CIMI, EBUM, ENCB, HUMO, IBUG, IEB, LL, MICH, MO, NY, SLPM, TEX, US, and XAL (herbaria acronyms according to Thiers, updated continuously). This data set was cross-checked with an unpublished list of known weeds in Mexico (Villaseñor & Espinosa, in preparation, an updated version of ^{Villaseñor & Espinosa, 1998}). This list was based on information from herbarium and literature sources, not all supported with specimens, with relatively broad and country-wide criteria. This process resulted in 426 species and 8,384 records of potential weeds in Michoacán. An additional search of the entire source database for habitat indications of either segetal or ruderal sites, but for species not found in the unpublished source list of weeds of Mexico, resulted in 4 more species and 37 records. Thus, the preliminary list of weeds of Michoacán consisted of 430 species and 8,421 records.

Figure 2 Diagram illustrating the selection of synanthropic Asteraceae species from Michoacán. 

Final list. The preliminary list was critically reviewed to confirm or reject occurrence in human-made habitats in Michoacán. A species was included in the Michoacán weed list if at least 2 independent sources of information confirmed the ability to grow in highly disturbed environments. These sources could either be herbarium labels with clear indications of disturbed habitat or the description of the habitat in regionally relevant Floras.

The habitat information of the following Floras was used: Flora del Bajío y de Regiones Adyacentes (^{Cabrera & Rzedowski, 2018}; ^{García & Koch, 1995}; ^{Redonda-Martínez, 2013}, ²⁰¹⁶; ^{Rzedowski et al., 2011}; ^{Rzedowski & Rzedowski, 1995}, ¹⁹⁹⁷, ²⁰⁰⁸; ^{Villarreal, 2003}; ^{Villarreal et al., 2006}), Flora Fanerogámica del Valle de México (^{Rzedowski et al., 2005}), Flora Novogaliciana (^{McVaugh, 1984}), the website Malezas de México (^{Vibrans, 2006} onwards), the work on Compositae in the synanthropic flora of Mexico (^{Rzedowski, 1993}), and some additional publications, such as Flora Mesoamericana (^{Pruski & Robinson, 2018}) and some taxonomic treatments of groups.

Habitat information from labels was extracted for the 8,421 records of the preliminary list from IBdata v3 (^{IBdata, 2021}). We excluded species that were only known as cultivated (Artemisia absinthium L. and Leucanthemum lacustre (Brot.) Samp.) or aquatic/subaquatic (Bidens laevis (L.) Britton, Sterns & Pogg., Jaegeria macrocephala Less., Leucosyris riparia (Kunth) Pruski & R.L. Hartm., Melampodium bibracteatum S. Watson, Solidago paniculata DC., and Tagetes persicifolius (Benth.) B.L. Turner), which resulted in 422 species and 8,360 records. Of these, only 6,539 records had habitat information. After an individual review of the label images on the IBData site, 1,739 additional indications of disturbed habitat were retrieved that had not been captured in the database.

The habitat information from the labels was classified into 2 types of environments. The first was predominantly natural or semi-natural (“NAT”) and the second was heavily disturbed (“WEED” for weeds). Semi-natural environments were included in the NAT category because disturbance is less intense and recurrent (^{Harlan & de Wet, 1965}), although they may be the sources of some weed species (^{Vibrans, 2002}), and generally have a composition of their own (^{Martorell et al., 2017}; ^{Rzedowski, 2006}). The natural or semi-natural category included all types of forest, mangroves, natural palm groves, shrublands, as well as grasslands if the label did not indicate they were induced or planted, and high-elevation grasslands. Sites described as logged forests or forests with disturbed vegetation, etc., were also considered semi-natural vegetation. Some examples of label information that were considered to refer to semi-natural vegetation were the following: “heavily logged forest 15 to 20 years ago or so”, “low deciduous forest, disturbed, with Bursera, Caesalpinia, Lonchocarpus”, or “hillside with oak forest vegetation, forest in disturbance, probably replacing pine forest”.

Some other open vegetation types were also included in (semi-) natural vegetation. Examples were clearings or forest edges, cliffs, rocks or lava flows (“pedregal”), canyons, riverbanks, shores, wetlands, or volcanic craters. Examples of descriptions of open vegetation included in the natural category were “clearings in the middle of Abies forest”, “at the edge of moist forest”, “Pinus and Abies forest, on rocks”, “tropical deciduous forest, on outcrops of igneous rock”, or “oak forest on lava flow”.

Plants growing in human-transformed and heavily disturbed habitats were considered weedy. They included habitats described as segetal (“arvense”), crops, plantations, ruderal, roadsides, dam banks, canals or ditches, field edges, near crops, reforestation areas, pastures or induced grassland, urbanized sites or near houses, wastelands, embankments, or secondary vegetation without indicating a secondary forest. A total of 3,192 different habitat descriptions were assigned to the 2 classes (Supplementary material 1).

We found at least 2 collection records from clearly disturbed sites for 265 species (Fig. 2). Another 60 species were added based on one label and additional bibliographic support for that habitat. Others (32 species) were added even without clear label information, but with at least 2 independent literature references indicating that they may behave as agrestal or ruderal weeds in adjacent regions. We discarded 65 species that had information from only 1 source (e.g., literature or a single label without clear information). We always found literature support for single collections from a clearly disturbed habitat.

For the resulting 357 weed species, we gathered the following information from the literature: tribe (^{Funk et al., 2009}), geographic distribution (native, but with distribution outside Mexico; endemic to Mexico; introduced), distribution by state in Mexico, geographic origin of the introduced species, as well as their life cycle and growth form (Appendix). Correlations were examined with linear regression analysis, and floristic similarity between states with the Sorensen-Dice similarity index, constructing a dendrogram with the UPGMA clustering method.

Results

From the preliminary list of 430 possible weed species of the Asteraceae, 73 (17%) were eliminated: 2 were only cultivated, 6 were aquatic or subaquatic, and 65 could not be documented as growing in disturbed environments in the state. As a result, 357 species were recognized as weeds for this state (Appendix).

Most synanthropic species also grew in less disturbed habitats; only 27% of all records were from clearly disturbed sites (Appendix), and a very low number, 2.7%, from crop fields (segetal weeds). Only 22% of the species (80; Appendix) grew mainly in disturbed sites and 16 species were found exclusively in disturbed sites. Of this number, 7 species had only 1 record, 5 species 2, 2 species 3 (Verbesina serrata Cav. and Conyza bonariensis (L.) Cronquist), and 1 species 5 records (Helminthoteca echioides (L.) Holub). Lactuca serriola L., an introduced species, was the most common weed growing exclusively in disturbed environments with 16 records.

Heliantheae had the highest number of weed species in Michoacán (83, Supplementary material 2); Liabeae (Sinclairia glabra (Hemsl.) Rydb.) and Mutisieae (Leibnitzia lyrata (D. Don) G.L. Nesom) had the least with only 1 species. For the state, no synanthropic species of the tribes Arctotideae, Calenduleae, Chaenactideae, Gochnatieae, Madieae, and Onoserideae were found. However, among the genera with the highest number of species, none belonged to the most species-rich tribe; the most outstanding were Melampodium L. (tribe Millerieae, with 19 species), Stevia Cav. (Eupatorieae, with 18 species), and Pseudognaphalium Kirp. (Gnaphalieae, with 14 species, Appendix).

The number of weed species per tribe correlated significantly with the total number of species of the tribe in Mexico (including non-weeds) (R² = 0.86, p = < 0.0001, Fig. 3). This correlation was even higher when comparing only species in Michoacán (R² = 0.89, p = < 0.0001). An exception was the tribe Eupatorieae with a relatively low proportion of weeds; this was also the tribe with the most species removed from the original list (26 species).

Figure 3 Relationship between the weedy and total number of species of Asteraceae by tribe in Michoacán and Mexico. The orange line and dots are for the data from Michoacán, and the blue line and dots compare weed data from Michoacán with all species known from Mexico. 

Native weed species (including endemics) dominated in Michoacán. Of the 357 accepted species, 342 (96%) were native and 150 (44%) endemic to Mexico. Heliantheae had the highest number of native and endemic species (83 and 42, respectively). Some tribes with relatively few species had a high proportion of endemic weeds, for example, Gnaphalieae and Cardueae (Table 1).

The pattern of endemic weeds was like that of all weeds, but they were found in fewer tribes (13). The correlation by tribe between the number of endemic weeds in Michoacán and all endemic species (weeds and non-weeds) in Mexico was high (R² = 0.77) and increased when only species endemic to Mexico and documented from Michoacán were considered (R² = 0.87).

We found 15 species of introduced or exotic weeds in the state (4.2% of all weedy species) distributed in 5 tribes (Table 2), all herbaceous. The tribes with the highest number of introduced species had few native species, particularly Cichorieae (5 genera, 6 introduced species, and only 5 native species). This tribe had only 1 species endemic to Mexico (Lactuca brachyrrhyncha Greenm.). Eleven of the 15 introduced species were native to Europe, and 7 of these were also distributed in Asia, and 3 in Africa. The South American species belonged to the tribe Anthemideae (Soliva anthemifolia (Juss.) Sweet and S. sessilis Ruiz & Pav.); Senecio inaequidens DC. came from Africa and Cotula australis (Spreng.) Hook. f. from Oceania.

Weed species native to Michoacán were shared mainly with neighboring states, particularly the Estado de México (324). States located in mountain ranges such as the Volcanic Belt and the Sierra Madre del Sur had the highest similarity (Figs. 4, 5). The number of shared species decreased towards the north; for example, Michoacán shared only 28 species with Baja California (Figs. 4, 5). Only a few species were known from all 32 federal states: Artemisia ludoviciana Nutt., Bidens pilosa L., Conyza canadensis (L.) Cronquist, and Symphyotrichum expansum (Poepp. ex Spreng.) G.L. Nesom.

Figure 4 Native weed species of Michoacán and their floristic similarity with other federal states of Mexico. Similarity index used: Sorensen-Dice, clustering method: UPGMA. 

Figure 5 Michoacán native weeds (endemic and non-endemic) shared with other states of Mexico. 

The Mexican endemic weed species showed a pattern similar to widespread native weeds. However, Guerrero and Oaxaca had less floristic similarity. Michoacán shared the highest number of endemic weeds with the Estado de México and Jalisco (136 and 135 species, respectively, Fig. 6). The endemic species with the widest distribution was recorded in 28 states (Stevia origanoides Kunth). The state with the lowest number of shared species was Quintana Roo (only Montanoa grandiflora Alamán ex DC., also recorded in 22 states).

Figure 6 Mexican endemic weeds of Michoacán shared with other Mexican states. 

Of the 150 weed species endemic to Mexico, 11 (3.4%) had a more restricted distribution and were limited to Michoacán and its neighboring states (Table 3). They belonged to 5 tribes, particularly Heliantheae. As with the endemic species, the closest relationship was with Jalisco (10 species) and Estado de México (6) (Fig. 7).

Figure 7 Mexican endemic weeds in Michoacán with restricted distribution and shared with neighboring states.  

The weeds of Michoacán were mainly herbaceous (79%) (Table 4), a little more than half of them perennial (56.9%). Annual herbs occupied second place, and a small part was biannual or behaved as both annual and perennial. A sizeable proportion was woody (19.3%).

Discussion

We found no previous work that separated weeds and non-weeds in a standardized and therefore reproducible way. This may be due to the intrinsic difficulties of such a distinction (^{Lema, 2017}). In nature, species live on gradients with different levels and types of adaptation to disturbance, both natural and human. These gradients make a clear separation between weeds and non-weeds difficult and will always be arbitrary to an extent; however, this arbitrariness is reduced with a clear definition of what is considered a weed. As mentioned in the introduction, here any species that could grow and establish populations in environments heavily disturbed by humans was considered a weed (^{Baker, 1974}; ^{Hanan & Vibrans, 2015}; ^{Rzedowski & Rzedowski, 2004}).

Other problems that had to be solved in a standardized way were lack of information, ambiguous information on herbarium labels or database records, as well as collection bias. Additionally, there were species in the gray or overlapping part of the gradient. These problems were solved by requiring 2 independent sources with information on their presence in heavily disturbed sites.

About one-sixth of the species cited as weeds in the literature from the region could not be confirmed as such. For example, some shrub species were part of pioneer vegetation and naturally disturbed sites but were not actually common components of transformed landscapes (^{Rzedowski, 2006}). Most of the species removed from the preliminary list were Eupatorieae, a tribe with numerous shrubby species that can grow along forest edges or roadsides, without being very weedy (^{García et al., 2011}).

A few excluded species were common weeds known from neighboring regions, but no specimens were found to document their presence in the state, perhaps due to insufficient collections. Examples are Ambrosia artemisiifolia L., Parthenium incanum Kunth, Grindelia subdecurrens DC. (^{Villaseñor and Espinosa, 1998}), or Ageratum conyzoides L. (^{Villaseñor, 2016}). The collection bias was also marked, as collectors generally avoid disturbed sites, especially cultivated areas (^{Carvajal, 1982}). However, although weeds (and especially arborescent weeds) are under-collected, the proportion of collections in disturbed environments does tend to reflect the degree of synanthropy (^{Hanan et al., 2016}).

The final list presented here contains species that can behave as weeds. It is probably incomplete, mainly because of insufficient data, a common problem with megadata (^{Hortal et al., 2015}). The number is likely to increase with better documentation. However, we suggest that our criteria (2 independent, unambiguous, and relevant records from heavily disturbed environments) can serve as a model for future work. Additionally, we urge collectors and databases to improve habitat information. Information on the general vegetation type of a site is useful, but disturbance level should also be included in herbarium labels and database fields.

In Michoacán, almost half (43%) of all Asteraceae were synanthropic. The proportion for the whole country is only 17% (^{Villaseñor, 2018}; Villaseñor & Espinosa ^{García, 1998}). This can be explained by differences in the size of the distribution areas -weedy species are much more widespread on average than non-weedy ones (^{Espinosa-García et al., 2004a}). More than half of the Michoacán weeds were registered nationally as weeds for the same reason (219 of 411; ^{Rzedowski, 1993}).

This study adds more than a hundred Asteraceae species of weeds to those previously reported for the state (^{Villaseñor & Espinosa, 1998}) and for all of Mexico (^{Rzedowski, 1993}) (132 and 138 more, respectively). Seven species considered by ^{Villaseñor and Espinosa (1998)} were not confirmed, although their presence was documented. They were not included because they lacked evidence for synanthropic behavior in the state.

There are no comparable data from other federal states of Mexico, except for Querétaro (^{Colmenero et al., 2001}). These authors reported 81 weed species of the family, that is, only a quarter of those recorded for Michoacán. Up to now, Michoacán is the state with the highest number of Asteraceae weeds in Mexico, confirming a statement by ^{Villaseñor and Espinosa (1998)} .

Few weed species (4.5%) were documented exclusively from heavily disturbed environments, and most of them with few records. Of these, Conyza bonariensis (L.) Cronquist and Simsia lagasciformis DC. had previously been shown to be restricted to disturbed environments (^{Rzedowski, 1993}). Only a quarter of the species had more than half of the collections in this type of environment. This partially reflected the above-mentioned collection bias towards more natural environments (^{Carvajal, 1982}). However, it also shows that in Mexico, ruderal and agrestal species are embedded in the vegetation of the adjacent landscape and depend at least partially on functional traits of species with non-weedy behavior (^{Bourgeois et al., 2019}; ^{Metcalfe et al., 2019}; ^{Munoz et al., 2020}).

Similar results have been reported in the Near East, where most weeds are also found in other vegetation types (^{Holzner, 1978}; ^{Zohary, 1973}). The results contrast with the more evident separation between weeds and non-weeds in central and northern Europe (^{Holzner, 1978}) or the United States (^{Sutherland, 2004}). This may perhaps be explained by the history of agriculture. In its region of origin, species preadapted to anthropized environments coevolved as crop weeds, then migrated along with the crops. The communities gradually impoverished but specialized, depending on environmental filters such as crop or soil type and climate (^{Fried et al., 2010}; ^{Smart et al., 2006}).

Generally, taxonomic groups are related, having characteristics in common as well as predictive value (^{McNeill, 1976}). So, if weeds are required to have certain characteristics (^{Baker, 1974}), one would expect them to be concentrated in certain tribes. However, the results of this study do not show such concentration.

The weedy species were distributed in 20 of the 26 tribes of Compositae known from Mexico (^{Villaseñor, 2018}); the 6 tribes without synanthropic species were species-poor. Generally, in Michoacán roughly half of the species in each tribe were weedy, though with some variation; endemic weeds also had a relatively constant proportion. The percentages at the national level were also proportional (range between 3.9% and 21.8%, with an average of 11.6%), but lower because of the more restricted distribution areas of non-weedy species on average (^{Espinosa-García et al., 2004a}). The only exception was Cichorieae, a tribe with few species but with a high proportion of weeds and exotics (^{Villaseñor, 2018}).

Some authors have proposed that weeds require specific characteristics to compete and colonize new environments (^{Baker, 1974}), which can, however, be highly variable (^{Basu et al., 2004}; ^{Hanan et al., 2016}). This study shows that weeds can develop in different tribes independently, perhaps because the functional spaces of weeds are not clearly delimited from non-weeds (^{Bourgeois et al., 2019}).

Under the criteria used in this work, the species-rich tribe Eupatorieae had a relatively low proportion of weeds. It consists mainly of perennial herbs or shrubs, often in forest edges or clearings (^{García et al., 2011}), where differentiation between weeds and non-weeds is particularly difficult. Also, the most species-rich tribes did not necessarily contain the most weed-rich genera; Melampodium (tribe Millerieae), for example, was the genus with the largest number of weedy species (19) but belonged to a medium-sized tribe; it grows widely in transformed landscapes at different levels of disturbance (^{Hanan et al., 2016}). ^{Rzedowski (1993)} highlights a group of incipient weed genera, which were indeed well represented as weeds in Michoacán, e.g., Stevia Cav. (18), Brickellia Raf. (9), and Erigeron L. (7).

In Michoacán, native weeds predominated (^{Rzedowski, 1991}). Almost half (44%) of them were endemic to Mexico (at a national level this percentage reaches 63.9% according to ^{Villaseñor [2018]}), but this was to be expected, given the generally larger distribution area of weeds.

The proportion of introduced species (4.2%) was much lower than that reported by ^{Espinosa-García et al. (2004b)} for the whole Mexican weed flora, with 28%. The differences are probably explained by differences in the proportion of introduced species between families (Brassicaceae, Fabaceae, or Poaceae have a considerably higher share of exotics; ^{Villaseñor & Magaña [2006]}). Most introduced species of the Asteraceae were native to Europe and the Mediterranean, which may be explained by the history of commerce (^{Villaseñor & Espinosa-García, 2004}). All were herbs, as occurs in other floras (^{Arianoutsou et al., 2010}; ^{Garcillán et al., 2013}; ^{Weber et al., 2008}), perhaps because this life form is well-represented in weeds and has some (pre-) adaptation to disturbed habitats.

The floristic composition reflects the history of biotic exchanges between regions and shared ecological conditions (^{Garcillán et al., 2013}). Comparisons provide information on possible factors that determine such similarities. The weed flora of Michoacán was most similar to that of neighboring states, a similarity that decreased as distance increased (^{Nekola & White, 1999}). However, we found stronger similarity with the western states located along the Volcanic Belt and the Sierra Madre del Sur (Guerrero and Oaxaca). They included regions known for their early settlements and agriculture (^{Zizumbo & Colunga, 2008}), high proportions of endemism (^{Estrada-Márquez et al., 2021}), and the greatest richness of Asteraceae (^{Villaseñor et al., 2005}).

There were relatively few (11) endemic weed species with restricted distribution, but it is interesting that they exist at all. ^{Rzedowski (2020)} in the Flora del Bajío y de regiones adyacentes highlights some of them as species restricted to the Balsas Depression (Chrysanthellum keilli B.L. Turner and C. michoacanum B.L. Turner) or to the Volcanic Belt (Verbesina pietatis McVaugh). ^{Estrada-Márquez et al. (2021)} consider Bidens pringlei Greenm. and Verbesina pietatis as restricted to the same areas. It is likely that these species are examples of relatively recently evolved weed species.

Weed species in Michoacán were mostly herbs (79%). Short life cycles facilitate growth in frequently disturbed open environments (^{Díaz et al., 2016}). However, a considerable number of woody plants could also grow in highly disturbed sites, as occurs in other parts of the world, such as France (^{Fried et al., 2009}). They should be examined in more detail.

In temperate regions, ruderals are mostly perennial hemicryptophytes or geophytes whereas agrestals are typically annual therophytes (^{Bourgeois et al., 2019}; ^{Holzner, 1978}), a pattern that was repeated here. The better representation of ruderal areas (roadsides, etc.) in the collections possibly increased the representation of perennials (63%), as cultivated areas (segetals) were severely under-collected.

Lack of quality information and collection biases limit the correct classification of species into weedy and non-weedy. However, herbarium data are good approximations when the information is reliable (^{Hanan et al., 2016}). This study is based on fully verified data, especially taxonomic identification. It is likely that other species that were not included due to lack of evidence will have to be added in the future.

This study proposes criteria to identify weedy species in a systematic, reproducible way and includes a species list based on these criteria. Michoacán harbors a high number of native and endemic Asteraceae weed species, with a very low proportion of introduced plants. Most species grow in both disturbed sites and natural vegetation. This indicates the existence of a wide variety of biological traits that allow plants to live in transformed environments, reflected also by the relatively similar proportion of weeds in the different tribes. Floristic similarities with neighboring states point to close relationships within the western and southern parts of Mexico, a region with a long history of disturbance associated with human settlements and agriculture.