What do monarchs feed on in winter? Nectar sources at hibernation sites

The steep population decline of the emblematic monarch butterfly is caused mainly by the reduction of food sources for caterpillars and adults, as well as disturbance in its overwintering forests. Although feeding at the overwintering sites in Mexico has long been considered unimportant, observations show that butterflies feed inside and outside of their forests on sunny days. Nectaring close to the hibernation colonies may be relevant for their conservation, as their reserves decline faster in disturbed forests. However, there are no systematic studies on nectar sources in the Monarch Butterfly Biosphere Reserve, Michoacán, Mexico. In this study, we identified and quantified the main plant species visited by butterflies for nectaring in the Reserve. We collected systematic observational data on the flora and butterfly visits in three sanctuaries, around colonies formed between February and March of the 2019–2020 season. Butterflies fed on 29 plant species from 10 families. Most had white, yellow or blue flowers, were somewhat synanthropic and had their main flowering season in winter. The most visited species were Salvia mexicana and S. plurispicata, which were also the most abundant. By individual plant, Prunus serotina, Crataegus mexicana, Buddleja sessiliflora, Verbesina oncophora and Roldana albonervia were the most visited. The results point to possible interventions in support of the butterflies. The visited species are generally easy to encourage or cultivate and could be promoted in the surrounding agricultural areas.


Introduction
The monarch butterfly, Danaus plexippus, is one of the most emblematic and beloved insect species in the world. This has several reasons: its well-known long-distance, multigeneration migration, its aposematism, that is the capture of toxic substances of plants as defense against predators, and its advertisement of unpalatability with a conspicuous color pattern (Malcolm 1994). However, it is under various Tithonia tubiformis (CONABIO 2014;SEMARNAT 2018). When the monarchs arrive at their hibernation sites, each individual has an average of 133mg of lipids, and at the end of the season it has 52mg (Alonso Mejía et al. 1995), also Alonso- Mejía et al. (1997).
Early studies stated that nectar consumption at the hibernation areas was unimportant because the numbers of butterflies far exceeded nectar sources (Brower et al. 1977). However, it is unclear if these studies also considered the surroundings of the hibernation sites. Feeding was sufficiently obvious for a group of investigators to study the difference in weight between feeding and non-feeding butterflies; they found that feeding butterflies were mostly severely underweight (Alonso- Mejía et al. 1997;Rendón Salinas 1997). However, they only studied the difference between butterflies nectaring within the forest and those that did not at that moment; apparently, they did not sample the very conspicuous "butterfly rivers" that form under appropriate circumstances (see below). Later, controversy erupted on the importance of winter feeding, with one author proposing to increase disturbance to promote nectar sources (see Hoth 1995) and the responses of Alonso- Mejía et al. (1995) and Brower (1995)). The reply of Brower emphasized the lack of information on these nectar sources and their relative significance but doubted their importance. Alonso-Mejía et al. focused on the importance of dense forests that help monarchs conserve energy, particularly under the occasional extreme weather conditions.
On sunny days -almost daily in the second half of their stay -monarch butterflies will form butterfly "streams" or "rivers" from their hibernation trees to lower-lying open areas; this has been interpreted as a search for water in order to process the lipids. Indeed, the butterflies seek out sources of moisture, but they also take advantage of flowering herbs, shrubs and trees to nectar. There are a number of flowering plants in the fir forests where monarchs hibernate, but even more so in the surrounding cultivated landscapes. Though most herbaceous species reproduce during the rainy season, a number of shrubs and trees of the area, as well as a few herbs, flower in the dry season, i.e., winter (Rzedowski 2006;Cornejo-Tenorio and Ibarra-ManrÍquez 2007). Rendon-Salinas (1997) found various species that may be attractive for the animals, belonging to such genera as Eupatorium, Senecio, Stevia and Salvia. So, nectar probably has a supportive function at least. Some other evidence points to the relevance of nectaring in the southern parts of the fall migration range (Brower et al. 2006;Hobson et al. 2020).
Butterflies consume their lipid reserves depending on their activity. Extremes in sunlight, relative humidity and temperature increase in disturbed or open forests. Thus, forest deterioration may deplete energy reserves in the monarchs, and result in fewer survivors with sufficient lipid reserves for mating and the flight back to their breeding grounds (Alonso Mejía et al. 1995;Brower et al. 2006;SEMARNAT 2018). Nectar sources near the hibernation sites could become more important under these circumstances (Brower et al. 2011).
Despite these observations, there are no systematic studies on the plants visited for nectaring from the Monarch Biosphere Reserve in Mexico. This study identifies the main feeding plants of monarch butterflies in their overwintering sites, separating the effect of plant species identity from species abundance. This information will be useful for forest management. It can also help to make the cultivated landscapes of the region friendlier for these animals by integrating the visited species into agroecosystems, or even for pollinator gardens as local attractions (Rogel-Fajardo et al. 2011;del Coro Arizmendi et al. 2020).

Study area
The Monarch Butterfly Biosphere Reserve, covering an area of 56,259 hectares with three core (13,551ha) and two buffer zones, is located between the States of Mexico and Michoacán, in Central Mexico. The three core zones are mostly situated above 3,000 m: Cerro Altamirano (588ha in the northern part of the reserve), Sierra Chincua-Campanario-Chivatí-Huacal (9,671ha in the center) and Cerro Pelón (3,339ha in the south). It has a cool to cold subhumid climate with summer rains (C(w2)(w)(b'(i"); SEMARNAT/ INECC 1996) and is part of the Trans-Mexican Neovolcanic Belt (Tamayo 1962), with volcanic rock subsoil. The forests of the Reserve are crucial for recharging the aquifers that supply groundwater to the Lerma-Santiago and Balsas hydrological basins, and urban centers such as Mexico City and Toluca (Gutiérrez Carbonell et al. 2001).
The vegetation is composed of oak, pine-oak, pine, fir, cedar and juniper forests, and grasslands (SEMARNAT/ INECC 1996); the monarchs hibernate mainly in the fir (Abies) forests. Villages and agriculture can be found at lower elevations. The reserve is home to at least 132 bird species, 56 mammals, 423 vascular plant species, and 211 fungi (Díaz Barriga 2002). Some endemic species such as the salamanders Ambystoma rivulare and Pseudoeurycea bellii (Parra-Olea et al. 2005) are found, as well as other protected species such as the Pinus martinezii and Juniperus monticola .
The large number of landowners makes conservation action complex (Gutiérrez Carbonell et al. 2001). Stakeholders include 59 ejidos and 13 Mazahua and Otomí indigenous communities; both communities and ejidos are forms of legal, communal land ownership in Mexico. There are 21 small properties and some state and federal-owned lands as well. Also, human development is an important goal, as the communities surrounding the Reserve are poor and lack employment outside of subsistence agriculture and seasonal tourism.
Forest degradation and loss of forest cover due to logging and land use changes are evident in parts of the Reserve (Brower et al. 2002). Natural factors such as rains, droughts and strong winds, exceeding the resistance threshold of these ecosystems, have deteriorated their structure and function in recent years (Vidal et al. 2014). An exceptional storm in 2016 substantially reduced the density of the fir canopy , and was followed by controversial salvage logging (Leverkus et al. 2017). The national and international increase in avocado consumption is a particularly strong incentive for land use change (Burnett 2016), though deforestation within the reserves appears to have declined considerably in the last decade due to better coordination of stakeholders and incentives for landowners (Flores-Martínez et al. 2020).
The investigation took place in a protected area that belongs to three ejidos. It was purely observational; we did not need a collection permit. The Reserve authorities were informed in writing of the project, as required. We visited each ejido president and explained the project. They assigned guides who supervised all field work and served as informants.

Selection of the study sites
In February to March 2020, during the period of this study, there were 22 hibernation sites in 15 sanctuaries (core areas) inside and outside of the Reserve (Rendón- . The colonies were located between 18°59'25" to 19°58'21" N and 98°41'37" to 100°49' W. The study sites were selected based on recommendations from local residents and staff from the World Wildlife Fund in charge of monitoring the hibernating monarch butterfly. They recommended working in sites with the largest butterfly colonies, considering occupation constancy and surface area within the forest: (1) Ejido "El Capulín", Cerro Pelón Sanctuary, municipality of Donato Guerra, State of México, south of the Reserve; (2) Ejido El Rosario, Sierra Campanario Sanctuary, municipality of Ocampo, State of Michoacán, center of the Reserve and (3) Ejido Cerro Prieto, Sierra Chincua Sanctuary, municipality of Angangueo, State of Michoacán, north of the Reserve. The most conserved site with the densest forest was El Rosario, Cierro Prieto was intermediate, and El Capulín had a relatively high level of disturbance and less dense fir forest (WWF-México 2020).
During the 2019-2020 season the monarch butterfly populations occupied a forest area of 2.83 hectares. The colony of the ejido El Rosario covered 1.27ha, Sierra Chincua 0.28ha in the ejido Cerro Prieto, and Cerro Pelón (Colonia de la Comunidad Indígena San Juan Xocunusco, the closest to the ejido El Capulín) 0.28ha (Rendón- ). In the 2020-2021 season the occupied area was 2.10ha, of which 0.73ha was located in El Rosario, 0.07ha in Cerro Pelón and there were no area records for the ejido Cerro Prieto (Rendón- Salinas et al. 2021).
To find the initial sampling locations, we asked knowledgeable local people, such as tour guides, about butterfly movements on sunny days. Most flight corridors were near streams, creeks and ravines, but also in some flat places. Sampling sites were placed close to the overwintering colony within the sanctuaries ("inside") and downstream of the flight corridors, outside of the sanctuary in sometimes more disturbed places ("outside").

Field methods
The feeding butterflies were observed on plots along transects for predetermined time periods (Matteucci and Colma 1982;Bautista Zúñiga et al. 2011); this is a common technique for documenting pollinators. In each of the three sanctuaries, we placed two transects with three observation plots each. For the first transect, the first 10 × 3m observation plot was located 100m from the colony and in the direction of the flight corridor. Another two observation plots were positioned at distances of 100m in the same direction (Fig. 1). Then, we searched for an appropriate and accessible site approximately 1km along the flight corridor, and again, three observation plots were placed there at 100m intervals. The size of the observation plots was based on pilot evaluations and considered the size of most flowering plants Fig. 1 Diagram of the sampling method. The orange circle represents the butterfly colonies and the green rectangles the observation plots their proboscis into the flowers. Each individual plant or 1 m 2 population (with a known number of individual plants in Salvia) was observed for 15s, moving clockwise. These 15-second observations were taken first for one species, then for the next. The butterfly numbers, plant species and individuals were documented with voice recording. Then, a photograph was taken of the whole plot. For each plot, the counts took about 30min; approximately 3.5h were needed for the two transects with three plots of each site. The count was interrupted if clouds or winds halted butterfly movement. The other sites were visited on different days.

Plants visited for feeding and their relative importance
We observed monarch butterflies feeding on 29 species which belonged to 10 botanical families (Fig. 2). The most common family was Asteraceae with 16 species, followed (mostly shrubs and some herbs), observable area, butterfly movement, floristic variation, and vegetation heterogeneity.
During the first visit, the plots were marked with string, and the coordinates noted. The observer then stood outside of the plot and watched butterfly movement. Flowering plant species on which the monarch butterflies posed and inserted their proboscis into the flowers were considered nectar sources or feeding plants. The individual plants of these species were numbered, measured (larger plants) or estimated based on a few measurements (dense populations of smaller plants, particularly Salvia): height and largest diameter, from leaf tip to leaf tip.
The species visited by the butterflies were photographed and identified with a field guide of the reserve (Cornejo-Tenorio and Ibarra-Manríquez 2008). Doubtful cases were photographed in more detail and identified with other floras. Formal botanical collections were not possible due to restrictions in the reserve.
The following variables and indices were calculated with the plant data (Matteucci and Colma 1982): • Frequency: the number of times one or more feeding plant individuals was found in the observation plots.
Presence was recorded in each plot, but the three observation plots of a transect were considered a unit for analysis. • Density: Number of plant individuals per plot.
• Cover. This refers to the proportion of the area occupied by the perpendicular projection of the aerial parts of the individuals of the species. The cover of the species was obtained by measuring the largest width of the sampled plant and using it as side length for a square. This overestimates the true cover, but as we used relative cover, it should not influence the order of relative importance adversely. From these data, the Importance Value Index was calculated with the Coefficient of Cottam and Curtis (1956), which simply adds relative frequency, relative density and relative cover. The highest possible value for this index is 300.
Later, the butterflies were counted directly during the central hours of the day (10.30-14.00). For the count, the observer stood quietly in the middle of the plot, facing north, and then counted or estimated the number of animals inserting Various Asteraceae with yellow and white flowers followed. The relative importance of the nectar-providing species in the vegetation determined frequency of butterfly nectaring (Fig. 5). The importance index had the closest relationship with the number of visits (R 2 = 0.78), with relative density a close second (R 2 = 0.72). Relative frequency and relative cover had lower values (0.65 and 0.48, respectively). However, a few species had more visitors than predicted by their importance, particularly Ageratina rivalis and Verbesina oncophora, which are weedy, attractive, winter-flowering shrubs.
Plant height did not correlate with butterfly numbers, when considering both average height per species and plant individuals (R 2 under 0.1).
by Lamiaceae with 4 species ( Table 1). Most of these species had white (10) or yellow (10) flowers, considering the spectrum of the human eye. An additional six species had blue flowers, two violet and one pink. Most species were somewhat synanthropic; some had populations within forests, but also in cultivated landscapes. Three were introduced (Agapanthus, Senecio inaequidens and Veronica persica). Human-perceptible scent did not appear to play a large role.
The species with most visitors were blue-flowered Salvia mexicana and S. plurispicata, as well as the white-flowered Ageratina rivalis (Fig. 3); however, this was related partially to their dominance in the vegetation (see Supplementary material 1). If visits per individual plant were considered (Fig. 4), two trees with white flowers (Prunus serotina and Crataegus mexicana) were in the lead, together with Buddleja sessiliflora; the latter has yellowish-beige flowers  As mentioned above, feeding visits concentrated heavily on a few groups of species, mainly composites: Eupatorieae, Senecioneae, Verbesina, blue-flowered Salvia, white-flowered Rosaceae trees and Buddleja. Although several other species flower at the same time, they contributed only 1.5% of the visits. Notably, the butterflies did not visit the red-colored flowers that are also abundant in the monarch fir forest, such as Salvia elegans Vahl. The same table

Dynamics of visits and differences between transects close and farther away from colonies
During the entire period, 5020 individual feeding visits were observed (Table 2). Figure 6 shows the distribution of butterfly feeding observations of all visits. The visits to the individual sites followed a similar pattern and are not shown here. terflies observed in all sites by species. The color of the bars represents the flower color of the species support mating that initiates between mid-February and the first days of March, after the end of the reproductive diapause (Alonso-Mejia and Arellano-Guillermo 1992). These authors also noted a slight weight gain of some females towards the end of the season, though they attributed it to the addition of spermatozoids.
The visited plant species were quite similar to those the butterflies use during summer. They were dominated particularly by white and yellow-flowered composites and some Lamiaceae (e.g., Rudolph et al. 2006 in Arkansas;Tooker et al. 2002 in Illinois). The study also confirms the qualitative observations of Rendón-Salinas (1997) on the preferred genera and flower colors. Orange and yellow are preferred innately by the monarchs (Blackiston et al. 2011; these experiments did not include white). Additionally, the butterflies fed on species that are assumed to be adapted to bee-pollination, such as blue-colored salvias. These species were not the favorites, but their local abundance made them important nectar sources. Monarch butterflies are known to adapt and learn about nectar sources (Blackiston et al. 2011).
It is possible that this feeding behavior has increased due to the worsening conditions of the fir forests or of the migration routes. Thus, winter feeding may buffer deteriorating conditions of other elements of the monarchs' livelihood. Alonso- Mejía et al. (1995) expressly stated that they did not see much feeding behavior in Mexico, but it is unclear whether their observations were systematic, on sunny days, or whether they included observations outside the forests. As mentioned above, feeding butterflies were smaller and with much lower lipid reserves than those that stayed in the colonies. The decreased canopy density of the fir forests also shows considerable differences between the sites as to species; however, all conform to these groups.
There were substantial differences between the relative frequency of flowering species in the transects closer to the colonies and those farther away (Supplementary material 1). Salvia mexicana and Ageratina petiolaris, both known weedy species, were more common in the outside transects; a number of other weedier species were found only in the more disturbed places.
We observed monarchs feeding on some species not found in the observation plots. These were Verbesina virgata, Ageratina mairetiana, Stevia subpubescens, Stevia ovata, Baccharis conferta, Senecio stoechadiformis, Salvia mocinoi, Lopezia racemosa, Clematis dioica, and Euphorbia pulcherrima, the pointsettia. On waysides outside of the reserve, butterflies were particularly attracted by Roldana albonervia for both feeding and sunning themselves. This is a large shrub with showy yellow-flowered inflorescences in corymbiform panicles; this species was present in the study plots but uncommon and stunted.

Discussion
Monarch butterflies clearly feed in and around their winter hibernation sites, contrary to some previous statements in the literature (e.g., Masters et al. 1988). On sunny days, the "streaming" butterflies visit both water sources and flowering plants in their surroundings.
This study was not focused on the proportion of feeding butterflies or on the contribution to their nutrition, but rather on the plants visited. This feeding, particularly during the second half of the winter stay in Mexico, may

Acknowledgements
We are grateful to the ejidatarios of the Santuarios El Rosario, Sierra Chincua and El Capulín, and particularly the people of the Paradero Macheros, for their patience, interest and help in obtaining the data. The biologist Eduardo Rendón-Salinas, Ing. César Ávalos, Lic. Guadalupe Gómez, Ing. Pablo Ángeles from WWF-Mexico, and Lic. Alejandra Hinojosa from DANAIDAS helped with technical advice. Ing. Amancio Consuelo-Isidoro assisted with the fieldwork and the photographs. The work was funded by grant #868068 of Conacyt (Consejo Nacional de Ciencia y Tecnología) for the master's thesis of the first author.
Author contributions All authors developed the concept of the study and the methodology, reviewed and edited the Spanish version of the manuscript. NST did the field work, data curation, analysis and wrote the first draft. JLPE provided project administration and supervision. HV provided supervision, assisted with plant identification and wrote the English version in collaboration with NST.

Data Availability
The data are available within the manuscript and in the Supplementary material.

Declarations
Declaration of competing interest The authors declare they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/. may also promote flowering plants in the understory which the stressed butterflies then visit.

Conclusion
The results point to possible interventions in support of the monarch populations: the promotion of the favorite food plants, particularly in the open areas around the Reserve. The somewhat weedy plants that could be encouraged in the agricultural areas surrounding the sanctuaries are Ageratina petiolaris, A. glabrata (which has an agreeable aroma), perhaps A. rivalis (though this is more of a forest plant), Verbesina oncophora, Roldana albonervia, Prunus serotina, Crataegus mexicana and the blue-colored salvias. These species would also combine pleasantly in color and size for pollinator gardens. There are likely limits to the possible support and buffer function of these plants, as they cannot substitute all other deteriorating elements of the butterflies' environment. We do not suggest further disturbance of the already stressed fir forests, but rather, promotion of a supportive landscape context. Future investigations should focus on the relative importance of winter feeding for the monarchs. Perhaps the nectar also contributes to water uptake. A repeat study on the lipid reserve differences between perched and nectaring butterflies, including "steaming" animals, would be useful, as would studies on nectar quantity and quality of the most-visited species, and their floral ecology. More should be known about the relationship between forest cover of the perching monarchs and their nectaring behavior, the relative importance of the forest understory and the surroundings, influence of management on both microclimate and nectar sources, the energy requirements of "streaming" and of reproduction, or possible dynamics of nectaring, which may surge toward the end of their stay or be directed towards certain plants.