Migration of Eastern North American monarch butterflies via the South-east and the Atlantic: evidence from stable isotopes, thin layer chromatography, DNA and phenotype

Abstract

Monarch butterflies, Danaus plexippus (L.), from eastern North America are well known for their incredible autumn migration to Mexico; however, not all monarchs follow this route. There is evidence of monarchs overwintering and reproducing in Florida, arriving to insular and continental Caribbean, and roosting by the thousands in the Yucatán, Mexico. This work aims to present evidence that these monarchs are part of two current migratory routes, that we consider as the south-eastern and the Atlantic routes, routes that were probably more prominent in the past. Monarchs were collected for 12 months in south Florida, for 4 years in Cuba in November and once in March, in the Yucatán and Guatemala at different times, and once in northern Venezuela. We used two independent techniques, stable hydrogen isotope (δ²H) analyses of the wings and/or thin layer chromatography (TLC), to trace the monarch’s natal grounds. We analysed the DNA of monarchs collected in St. Marks in the Florida Panhandle, Cuba and Guatemala, and compared those data against previously-analysed DNA data from monarchs in the Americas to characterize their genetic structure and to assess the possible movement and presence of North American monarchs and/or their alleles outside the USA. Our results support the existence of south-eastern and Atlantic migratory routes. TLC, isotope and DNA analyses showed the arrival of likely North American monarchs in Cuba, Yucatán, Guatemala, Venezuela and other areas of the Americas. North American monarchs found in these areas have different natal grounds, phenotypic traits and DNA signature than Mexican migrants. Monarchs from the south-eastern route mostly originated in the south-east USA and fed on local Asclepias spp., such as Asclepias viridis, Asclepias humistrata, Asclepias perennis and Asclepias asperula. Butterflies from this migratory route move east, enter the Florida Peninsula, pass to Cuba, fly to the Yucatán and then to Guatemala where they appear to overwinter in the high mountains of Guatemala where Abies guatemalensis occurs. Monarchs that are part of the Atlantic route move east of the Appalachians, enter the Florida Peninsula, and from there pass to the insular and continental Caribbean. The main host plants for the Atlantic monarchs are A. perennis and A. humistrata; in contrast, Mexican monarchs mainly feed on Asclepias syriaca. Some monarchs from these two proposed migratory routes, south-eastern and Atlantic, will stay in the places where they travelled and others will return via Florida and Mexico. We propose a scenario for how the different migratory routes evolved.

Atlantic, Caribbean, DNA, Guatemala, isotopes, migration, Monarch butterfly, phenotype, south-eastern, TLC

INTRODUCTION

The monarch or milkweed butterfly (Danaus plexippus Linnaeus, 1758) is present in the Americas from most of North America to the northern part of the Amazonas (Williams et al., 1942). Along its distribution several subspecies are described, Danaus plexippus plexippus (Linnaeus, 1758) , is present in North America and some authors like Godman & Salvin (1879–1901) describe its presence until Nicaragua. Danaus plexippus megalippe Hübner, 1826, the second-most abundant subspecies, occurs from the southern part of North America, Central America to northern South America, as well as in the Caribbean (Williams et al., 1942). Besides D. p. megalippe there are several subspecies that are recognized in the Caribbean, Danaus plexippus portoricensis (Clark, 1941), among others. But the most iconic of them is without doubt, D. p. plexippus, the only subspecies known to migrate. The generally accepted view is that there are two migrant monarch populations in North America, west and east of the Rocky Mountains (Brower, 1985; Urquhart, 1987). The population that occurs west of the Rocky Mountains breeds in much of the western USA and southern British Columbia in Canada, and migrates short distances to several overwintering locations along the Pacific coast in patches of eucalyptus (Eucalyptus spp.), Monterey pine (Pinus radiata D. Don) and Monterey cypress [Hesperocyparis macrocarpa (Hartw.) Bartel.] (https://www.fs.usda.gov/wildflowers/pollinators/Monarch_Butterfly/migration/index.shtml?fbclid=IwAR17cqz8_tHTDC_c5kg422hr4oe0ZohyoH7Z2tjMY2H7WDV56iEZxwvyF7A#:~:text=Overwintering%20in%20Mexico&text=Monarchs%20roost%20for%20the%20winter,2%20miles%20above%20sea%20level). The eastern population migrates thousands of kilometres from the Midwest and surrounding region to overwinter in the high elevation oyamel fir [Abies religiosa (Kunth) Schltdl. & Cham] forests in Mexico. Several accepted migratory routes or pathways are followed by the eastern monarchs, the main one goes through the eastern plains in North America, another runs along the western portion of the Panhandle, and the last runs along the Atlantic coast including the Florida Peninsula. This last route does not appear to take migrants to the Mexican overwintering colonies (Monarch Watch, 2016) (https://monarchwatch.org/tagging/index.html#recoveries).

Monarchs that arrive in the high altitude oyamel forest in Mexico to overwinter stay inactive and in reproductive diapause for several months. Reproductive diapause allows the monarchs to delay reproduction until early spring when they mate and migrate back to the southern USA to exploit freshly emerging milkweed plants (Herman, 1985; Malcolm et al., 1993). Three to four generations of monarchs are needed to arrive to the Midwest to reproduce (Malcolm et al., 1993). The offspring of the monarchs that arrive to the Midwest to reproduce migrate south to the Mexican overwintering grounds, thus resuming the migratory cycle.

Before the 1975 discovery of the overwintering colonies in Michoacán, Mexico, by Urquhart and his collaborators (Urquhart, 1976), there were reports of monarchs that did not fit the two monarch migratory population scenarios, including their routes, migration times, phenotypic traits and behaviour. Some of these reports indicated monarchs arriving at Englewood in western Florida, USA, in late October and November in the 1930s (Hodges, 1937 in Williams et al., 1942) and monarchs roosting in the autumn in central Florida (Urquhart, 1960; Urquhart & Urquhart, 1976; Brower, 1995). South of the Florida Peninsula, tagged migrant monarchs were found in Miami and the Florida Keys, probably Key West (Urquhart & Urquhart, 1976), and 19 monarchs that were tagged in northern USA states and one in Ontario, Canada, were recaptured in Florida during the autumn and early winter (http://www.monarchwatch.org/tagmig/recoveries.htm).

Knowledge of migrant monarchs south of the Florida Keys into the continental and insular Caribbean is tenuous since migrant monarchs in these areas are considered part of an ‘aberrant migration’ (Urquhart, 1987) or a ‘dispersal route’ (Brower, 1995). Throughout this paper, the term ‘migration’ is defined behaviourally and physiologically sensuHobson et al. (2019b) as ‘… the movement away from the home range that does not cease [at least not initially when suitable resources are encountered] which requires a set of behavioral and physiological adaptations for sustained movement that are unique from day-to-day adaptations related to self-maintenance and breeding …’. In contrast, dispersion or ranging is ‘… to separate and become further apart’ (Taylor, 1986), it is a ‘one way movement that takes the animal away from the animal’s natal grounds’ (Dingle, 2014).

Evidence of migrant monarchs south of the Florida Keys into the continental Caribbean is sporadic but compelling. One of the first records of migratory monarchs reaching the Caribbean was an individual tagged in Ontario (Urquhart, 1979b) and recaptured in Havana, Cuba (L. de Armas, pers. comm.). This report was followed by other recaptures from Urquhart’s (1987) 50-year-long tagging programme in North America: two in Hispaniola, one in Jamaica, one in Puerto Rico, one in the Lesser Antilles and one in Trinidad (only a few miles from the South American continent). Over a 3-year period, Dockx (2002) found migrant monarchs arriving annually to western Cuba (the eastern side was not sampled) in November where migrants outnumbered resident monarchs, and most migrants were reproductively active. In the westernmost portion of Cuba, near Cabo San Antonio, Dockx (2002) observed that, when winds were blowing towards the Yucatán Peninsula (Fig. 1), monarchs departed in the same direction across the Yucatán Channel. In the north of Isla Mujeres (Fig. 2), 161 km away in the Yucatán, monarchs have been observed flying over the sea (J. Flores, pers. comm.; Supporting Information, Fig. S1). The flight direction when arriving on the Yucatán Peninsula was southward. Monarchs arrived at the end of the year in the Yucatán Peninsula in 1977 and 1978 and were observed roosting (B. MacKinnon & G. Cobb, pers. obs.) suggesting that they were migrants. In mid-October 1977, thousands of monarchs were observed feeding on flowers and hanging from palm trees in the northern section of Celestún (Fig. 2), on the western coast of Yucatán (B. MacKinnon, pers. obs.). A video showed the arrival of thousands of monarchs in Celestún in late October and early November 2007; the monarchs were clinging to pine needles where they spent the night before continuing their journey (https://www.youtube.com/user/Yucamama). Observations in the past 5 years confirm the continued arrivals of likely, migratory monarchs in the Yucatán Peninsula that, in some cases, formed aggregations of around 300 individuals (Supporting Information, Fig. S2 and S3). Urquhart & Urquhart (1979b) recaptured four Canadian tagged monarchs in the Yucatán: one in the Celestún area and the remaining three along the East Coast.

Figure 1.

States and countries with historical monarch observations supporting a south-eastern and Atlantic migratory route, and the main Mexican overwintering site for eastern monarchs. The numbers show observations of monarchs, see Supporting Information, Table S1.

Figure 2.

Collection sites (★) and/or observations (•) of monarch butterflies in Yucatán, Florida and Cuba.

What happens to the monarchs that arrive in the Yucatán Peninsula? Mexican migrant monarchs overwinter in high altitude oyamel forests that allow them to conserve energy during the long winter, but there are no mountains in the Yucatán. We know from MacKinnon, Cobb and the video (https://www.youtube.com/user/Yucamama) that the thousands observed roosting dispersed after a few days (B. MacKinnon, pers. comm.). Observations by MacKinnon and Urquhart during October 1977 and 1978 also indicated that the monarchs that arrived on the west and north west coast of the Yucatán Peninsula moved north-east along the coast before moving south along the east coast (Urquhart & Urquhart, 1979b). Urquhart & Urquhart (1976) reported three migrant monarchs in Guatemala, plus monarchs overwintering in the Sierra Madre Mountains of Guatemala and western Honduras (Urquhart & Urquhart, 1979b; Fig. 1). They were unable to continue research on overwintering colonies in Guatemala due to political unrest (letter to B. MacKinnon).

Monarch migration routes and natal grounds have been studied using four main methods: mark-recapture (i.e. tagging), thin layer chromatography (TLC), stable isotopes and, more recently, DNA. Tagging of monarchs began in the 1960s and played a decisive role in understanding the monarch’s migration routes and the discovery of their overwintering colonies in Mexico (Urquhart, 1987). However, tagging requires significant effort over long periods and recovery rates of tagged monarchs are very low: only 56 of 11 333 (0.49%) individuals tagged in southern Pennsylvania were recovered at the Mexican overwintering sites (Steffy, 2015), and Monarch Watch reported a recovery rate around 1.18% ((Monarch Watch, 2016). We propose that the low recovery rate reported by Steffy (2015) could be the result of some monarchs entering the Florida Peninsula.

TLC (Roeske et al., 1975) can potentially identify the general natal ground of an individual monarch butterfly from the cardenolide fingerprint pattern unique to each milkweed (Asclepias) species, the monarch host plant that the larval butterfly consumed. By matching the TLC fingerprint with the host plant species’ geographic distribution, the general area where the butterfly hatched can be determined (Malcolm et al., 1993). However, only a fraction of TLC profiles exist for the > 100 native and non-native milkweeds in North America (https://monarchjointventure.org/images/uploads/documents/monarchsandmilkweed_may29.pdf), some species have similar profiles and others like A. syriaca have significant intra-fingerprint variation (Malcolm et al., 1989) making the assignment to a particular Asclepias sp. subjective and potentially confounding identifications.

The stable hydrogen isotope (δ²H) of monarch wings has been instrumental in determining the natal grounds of monarchs, including individuals from overwintering colonies in Mexico (Wassenaar & Hobson, 1998; Flockhart et al., 2017; Hobson et al., 2019a). As a result, we have reliable isotopic base maps (isoscapes) for assigning δ²H isotope values for the east of the Rockies monarch populations around a latitude of 33°N (Wassenaar & Hobson, 1998; Hobson et al., 2019a), but further south of this geographical area, the isotopic values are either not well known or are influenced by a maritime effect (Clark & Fritz, 1997), limiting their use to study monarch movement (Contina et al., 2022). More recently, sequencing hundreds of monarch genomes is helping to trace the historical and contemporary genetic signature of D. plexippus and to determine this species’ possible geographical origin, movement and the unique DNA signatures for monarch populations around the world (Zhan et al., 2014; Talla et al., 2020). Combining multiple endogenous markers could potentially increase accuracy and precision when identifying natal grounds of individuals and butterfly populations.

The objectives of this study were to determine the potential existence of south-eastern and Atlantic migratory routes and to show that each have distinct natal grounds, migratory routes and phenology, and their own genetic signature and phenotype. The northern boundary of the south-eastern migratory route roughly comprises the area south of parallel 36°30’ and its western boundary runs through Arkansas and Louisiana, and the Atlantic route’s boundary comprises the coastal region east of the Appalachians, including Florida. To test the existence of these two migratory routes, south-eastern and Atlantic migratory, we identified natal grounds and possible migration routes of monarch specimens collected in Florida, Cuba, the Yucatán Peninsula, Guatemala and Venezuela (Fig. 1). To accomplish this, we used multiple approaches including TLC, wing δ²H_w values and genetic signatures of monarch specimens. We postulate that monarchs that follow this south-eastern migration route would mostly originate in this area and fly south through the Florida Peninsula, cross to Cuba, continue on to the Yucatán Peninsula, and winter in Guatemala. Monarchs from the Atlantic route will mostly originate from the Atlantic flyway east of the Appalachians, and will migrate south through the Florida Peninsula into the insular Caribbean and, in some instances could reach northern South America. We suggest these monarchs are not part of an ‘aberrant migration’ (Urquhart, 1987,) nor a ‘dispersal route’ (Brower, 1995) but instead represent distinct migratory routes, with their unique phenotypes and genotypes. DNA evidence plus natural history observations support the North American monarchs return and indicate that they are not an evolutionary sinkhole.

MATERIAL AND METHODS

The main objective of this paper was to explore the existence of south-eastern and Atlantic migratory routes and their associated phenotype and genotype. Since these routes pass through North America and go beyond, we surveyed the presence, phenotype and genetic signature of the North American monarchs, D. p. plexippus, in the south-eastern and Atlantic regions of the USA and in local monarch populations, and D. p. megalippe, in the insular and continental Caribbean, Mexico, Guatemala and Venezuela where the subspecies D. p. megalippe is present. Monarchs in Louisiana, Georgia, Florida, USA, Cuba and Venezuela were collected in the field, and monarchs from Mexico and Guatemala came from collections. The monarch natal grounds were determined through TLC and/or stable isotopes (two independent techniques) (Table 1). In addition, we used DNA analysis to investigate the possible movement and introgression of North America monarchs and/or their alleles into the Neotropics. Table 1 summarizes the samples, procedures and associated tables and/or figures.

Table 1.

Summary of the butterfly samples, procedures and/or analyses performed. The following abbreviations were used: TLC (thin layer chromatography); DNA analyses; Field vs. collections indicates the source of the specimens; Extended Data (E.D.) Tab (Supporting Information, Table)

Sample	N	Date(s)	TLC	Isotope (δ2Hw)	DNA	Field vs. collection
Louisiana, USA	30	16 Apr.–3 May 1996	Table 2			Field
Georgia, USA	1	* Apr. 1996	Table 2			Field
Florida other than Miami§	317	6 Apr. 1987–20 Nov. 1995	Table 2			Field
Miami, Florida§	411	* Apr. 1990, * Sep. 1994, Nov. 1995	Table 3			Field
St. Marks, Florida, USA	21	2–4 Dec. 2015		Fig. 3	Figs 11–13	Field
Cubaǂ	138	10–25 Nov. 1993, 1995–1997,10–15 Mar. 1995	Table 4	Figs 5–7	Figs 11–13	Field
Yucatán, Mexico	28	1 Nov. 1990–20 Jul. 1999		Fig. 8		ECOSUR Collection
Catemaco, Mexico. Collected by Welling	4	20 Sep. 1964	Table 5			Former Allyn Collection
Colima, Mexico. Collected by Welling	1	15 Sep. 1975	Table 5			Former Allyn Collection
Guatemala. Collected by Welling	20	26 Sep. 1965–4 Oct. 1966	Table 5			Former Allyn Collection
Guatemala, U.Valle	11	1 Feb. 1978–* Dec. 2007		Fig. 10	Figs 11–13	Arthropod Collection
Guaranito Town, Portuguesa State, Venezuela.	4	27 Oct.–2 Nov. 1995	Fig. 4, E.D. Table 2			Field

* Unknown date. ^§ and ǂ data sets have been partially or totally published (Knight, 1998 and Dockx et al., 2004) but their TLC classifications have been redone.

Sample collections

Monarchs were collected in the field in Louisiana and Georgia, at different locations, and in the Florida Peninsula, two locations in Cuba in November and one location in Venezuela (sampled between October and November, 1995) (Figs 1–2; Table 1). In Mexico, E.C. Welling collected monarchs at two locations, in Catemaco, located in southern Veracruz, in September 1964; and, in the state of Colima, in September 1975 (Table 1; Fig. 9; Newcomer, 1967). In Guatemala, Welling collected in September and November 1965 in Chimaltenango, and from June to October 1966 in Alta Verapaz (Table 1; Fig. 9; Newcomer, 1967).

In addition, we analysed 28 monarchs collected in the Yucatán Peninsula that were part of the ECOSUR collection and nine in Guatemala that were part of the arthropod collection of the Universidad del Valle de Guatemala. Only specimens with characteristics of the migratory subspecies, D. p. plexippus (Dockx, 2012), were selected from these two collections, independent of the month when they were collected. Figure 1 shows the general areas where the monarchs were collected. In Venezuela, monarchs were collected in a grassy area around the town of Guanare in Portuguesa state (Fig. 1; Table 1; Supporting Information, Table S2).

Thin layer chromatography (TLC)

Cardenolides were ethanol-extracted from the de-fatted butterfly material in 10 mL volumetric flasks and TLC was performed on the cleaned samples following Brower et al. (1982) and Malcolm et al. (1989). Digitoxin and digitoxigenin cardenolide standards (10 mg) were spotted on the centre and sides of each plate. After the TLC plate dried, it was photographed and transferred digitally to visualize each TLC fingerprint pattern. These fingerprints reflect the milkweed host plant upon which that monarch fed as a caterpillar. We determined the host plant species by comparing spot intensity and mobility with known fingerprints of Asclepias asperula (Decne) Woodson (Martin & Lynch, 1988); Asclepias humistrata Walter (Martin et al., 1992); Asclepias syriaca L., (Malcolm et al., 1989); Asclepias perennis Walter (Moranz & Brower, 1998); Asclepias viridis Walter (Lynch & Martin, 1987); and Asclepias incarnata L. (L. P. Brower, 1991), or with fingerprints of monarchs reared on Asclepias curassavica (L) Kuntze (Dockx 2002). Most of these Asclepias spp. grow predominantly in the south-eastern USA.

Results from the TLC analysis were used to match the Asclepias sp. signature from individual monarchs to the geographic distribution of the Asclepias sp. to obtain a general area where individual monarchs could have eclosed. Adults collected outside the USA that had a larval host plant range not present in the areas of collection were classified as migratory. A complete description, distribution map, photographs, and references of many Asclepias spp, including those used to classify host plant species in this study, are provided by the Missouri Botanical Garden website (http://www.mobot.org/) and the USDA (http://www.plants.usda.gov). The initial TLC fingerprints interpretations of Florida (Knight, 1998,) and Cuba (Dockx et al., 2004,) monarchs were carried out in the Lincoln Brower laboratory, only two TLC fingerprints were identified, A. syriaca and A. curassavica. Years later, observing the TLC fingerprints from the monarchs collected in Guatemala, Florida and Cuba (Table 1), distinct fingerprints that were not the initial A. curassavica and/or A. syriaca were observed in these three samples, prompting their reanalysis. Around 13% of the A. syriaca fingerprint inKnight (1998,) andDockx et al. (2004,) and 45.5% of curassavica and/or undetermined fingerprints were reclassified from the initial classification to Asclepias spp. that grow in the south-east USA. A blind reclassification of the TLC fingerprints of Florida and Cuba was done using published and unpublished TLC fingerprints to identify the Asclepias spp. The verification of A. asperula, A. viridis and A. humistrata was carried out by two of the authors, Lynch and Dockx, Lynch was a co-author of the papers describing the fingerprints of these three Asclepias spp., monarchs with A. perennis fingerprints were sent to Moranz for verification (Moranz & Brower, 1998).

Stable hydrogen (H) isotopes

Monarch wing samples were cleaned and rinsed in a 2:1 chloroform: methanol solution and air dried before stable isotope analysis. Subsamples were cut from the same region of the hindwing to reduce inter-sample variance due to isotopic effects from pigmentation (Hobson et al., 2017), weighed (0.35 ± 0.02 mg) and encapsulated into silver capsules. Samples were prepared for δ²H analysis at the Stable Isotope Laboratory of Environment and Climate Change Canada, Saskatoon, Canada. All measurements were performed on a high temperature combustion (HTC) unit (Thermo Finnigan, Bremen, Germany) equipped with a Costech Zero-Blank auto sampler (Thermo, Thermo Instruments, Bremen, Germany ). The helium carrier gas rate was set to 120 mL/min through a 0.6 m ¼-inch 5-Å molecular sieve (80–100 mesh) GC column. The HTC glassy carbon reactor was operated at 1400 °C and the gas chromatoraphy (GC) column at 90 °C. After separation, the gases were introduced into a Delta V plus isotope-ratio mass spectrometer via a ConFlo IV interface (Thermo Finnigan). We used keratin reference standards CBS (Environment Canada Caribou Hoof Standard) and KHS (Environemnt Canada Kudu Horn Standard) to calibrate sample δ²H (-197 and -54.1‰, respectively; Qi & Coplen, 2011). These standard values were used instead of the newly reported values of Soto et al., (2017) to maintain consistency with our earlier published work for monarch assignments. This does not affect the assignment per se (providing appropriate calibration algorithms are used, see below). Based on replicate (N = 5) within-run measurements of keratin standards, sample measurement error was estimated at ± 2‰ for δ²H. All δ²H results are reported for non-exchangeable H in typical delta (δ) notation, in units of per mil (‰), and normalized on the Vienna Standard Mean Ocean Water-Standard Light Antarctic Precipitation (VSMOW-SLAP) standard scale.

Probabilistic assignment to origin

We used a probabilistic assignment to origin method to estimate natal origins of monarchs sampled in different regions using the assignR package v.2.2.1 (Ma et al., 2020) in the R computing environment v.4.1.1. Using this package, we created monarch wing δ²H_w isoscapes based on a transfer function derived from wing stable isotope values of known origin monarchs from Hobson et al., (2019a) by calibrating amount-weighted growing-season mean precipitation δ²H (δ²H_p) (Bowen et al., 2005) surface into a δ²H_w isoscape. This resulted in a calibration equation to a monarch wing isoscape of: −73.28 + 0.91 * δ²H_p, which is similar to the calibration equation of Hobson et al. (2019a). We then depicted potential origins for each monarch using probabilistic assignment techniques as described in Ma et al. (2020). In brief, the assignR package uses a likelihood-based assignment method similar to approaches developed by others (Hobson et al., 2009; Wunder, 2010; Van Wilgenburg et al., 2012) to assign individual monarchs using δ²H_w to the calibrated wing isoscape. However, the assignR method propagates error in the assignments via another spatial surface representing precipitation isoscape uncertainty (Ma et al., 2020). We included monarchs in the assignment analyses which had δ²H values < -90‰ (VSMOW) and/or TLC fingerprints from North America, as their values and fingerprints beyond this area are not well known and the main purpose of this work is to explore the movement of eastern North American monarchs past this region. These assignments use the current known geographic range for the eastern monarch population and also included Cuba and the Yucatán Peninsula as a spatial mask (i.e. clipped) to represent the most plausible regions of origin. Likelihood of a cell (pixel) within the δ²H_w isoscape representing a potential origin for a sample was estimated using a normal probability density function (pdf) based on the observed δ²H and thus depicted the likely origins of each monarch by separately assigning individuals to the δ²H_w isoscapes, separately. This resulted in individual probability maps for each assigned individual. Our spatially-explicit depictions should be considered reasonably informative for latitude, but rather ambiguous for longitude based on the structure of the underlying δ²H_w isoscape. Regions showing highest probability on these assignment maps relate to origins consistent with the isotope data. Geographic assignments to origin and spatial data operations were conducted using functions within the R statistical computing environment (R Core Team, 2022) with the ‘assignR’ (Ma et al., 2020), ‘raster’ (Hijmans, 2020), ‘rnaturalearth’ (South, 2017) and ‘rgdal’ Bivand and Lewin-Koh (2020) packages.

DNA analyses

To characterize the population structure and genetic history of monarchs in the Americas, after the work of Zhan et al., (2014), we newly sequenced 21 monarchs collected in St. Marks, Florida (in December 2015), six from Cuba and five from Guatemala. These individuals were added to the six individuals collected in the same St. Marks locality in October by Zhan et al., (2014). We combined the genotype data of these three groups of monarchs with sequence data from samples described in Zhan et al., (2014), which were sourced from populations across the Americas: North America (Mexico, California, Texas, Massachusetts, New Jersey, north and south Florida) and Central/South America (Aruba, Belize, Bermuda, Costa Rica, Ecuador). We excluded Atlantic and Pacific dispersal populations from the analysis in order to increase power to detect discriminant variation among North, Central and South American samples.

Genomic DNA analyses were performed using the abdomen and/or thorax from St. Marks and Guatemalan monarchs and chemical left-over material for the Cuban monarchs. A total of 1 µg of genomic DNA from the Guatemala individuals was fragmented to 180–220 bp using a Covaris S1 series sonicator. The DNA from the Cuban specimens was not fragmented as the extracted DNA was degraded and of low quality. The sequencing library was prepared with the fragmented DNA using a KAPA Hyper Prep Kit (KR0961, v.1.14, Roche Diagnostics, Indianapolis, IN) and sequenced on an Illumina HiSeq 4000 over three lanes to generate 100 bp paired end reads. Altogether, sequencing consisted of 284.3 Gb of new sequencing data (21 St. Marks samples, mean genome coverage: 18.5 ×; six Cuba samples, mean genome coverage: 5.2 ×; five Guatemala samples, mean genome coverage: 5.1 ×). Sequencing was carried out at the University of Chicago Genomics Facility.

Raw sequence data were mapped to the monarch butterfly reference genome (v.3, repeat masked) (Zhan and Reppert, 2013) using the Burrows–Wheeler Alignment algorithm (Li et al., 2009). Read duplicates and improper pairs were sorted and filtered using SAMtools 1.3.1 (Li, 2011) and Picard 2.4.1 (Picard Toolkit. 2019. Broad Institute, GitHub Repository. https://broadinstitute.github.io/picard/; Broad Institute). Genetic structure and admixture among samples were assessed in ngsAdmix (Skotte et al., 2013) using estimated genotype likelihoods computed in ANGSD (Korneliussen et al., 2014). Individuals with poor sequence coverage (genotyping rate < 0.1) were excluded from analysis prior to genotype likelihood determination, yielding 82 total individuals in the final analyses (all newly sequenced individuals from St. Marks, Florida, Cuba and Guatemala were included). The following parameter settings were changed from default in ngsAdmix: -misTol 0.95 (tolerance for considering a site as missing) -minLrt 0.95 (minimum likelihood ratio value for maf>0) -minInd 76 (minumum number of informative individuals), yielding 825 074 SNPs for final analyses. Principal component analysis was performed in PCAngsd (Cheng et al., 2017; Meisner & Albrechtsen (2018)) using the identical SNP set. General relationships as previously identified (Zhan et al., 2014) were recovered using these parameters.

Phylogenetic relationships between individuals were inferred from a maximum likelihood phylogeny produced in RAxML v.8.2.12 (Stamatakis et al., 2014) with 200 bootstrap replicates. Genotypes were called using the HaplotypeCaller and GenotypeGVCFs tools of GATK v.4.0.0 (McKenna et al., 2010). Default parameters were used except for increased heterozygosity prior (0.05), minimum base quality (30) and variant calling confidence (20). For computational efficiency, SNPs were restricted to homozygous variants called in a minimum of 64 (of 82) individuals. Alignments for analysis were prepared using VCF-kit (Cook & Andersen, 2017).

RESULTS

TLC and stable H isotopes

Florida, Louisiana and Georgia

In the spring, in Louisiana the majority (28 of 30) of monarchs collected were males, 19 with A. syriaca and two with A. viridis fingerprints, and in Georgia only one male had the A. viridis fingerprint (Table 2). Most monarchs sampled in the Florida Peninsula (Fig. 2) in the autumn through to the spring had the A. syriaca fingerprint, followed predominantly by Asclepias species with ranges in the south-east: A. viridis, A. exaltata and A. perennis (Table 2). A few monarchs with the A. curassavica fingerprint, an introduced species, also occurred in the Florida Peninsula. In the spring, in a northern Florida location known as Cross Creek (Fig. 2), similar TLC patterns to the other Florida monarchs were present: monarchs with the A. syriaca fingerprint were the most abundant group, followed by A. viridis, A. curassavica, and a few individuals with A. exaltata, A. humistrata, A. asperula capricornu or A. perennis fingerprints.

Table 2.

TLC fingerprints of monarchs collected in Louisiana, Georgia and different Florida locations during the 1980s and 1990s in the autumn and the spring. The fingerprints include A. humistrata (A.hu), A. curassavica (A.cu), A. asperula capricornu (A.as), A. viridis (A.vi), A. perennis (A.pe), A. syriaca (A.sy), A. exaltata (A.ex) plus undetermined fingerprints (Und.). The first number in each cell shows the number of females followed by males with that particular fingerprint and when there is a single number the individual is/are a female/s. See Supporting Information, Figure S4, for a composite TLC plate with the previous Asclepias spp. fingerprints. Specific dates are not available (*) for samples collected in January 1988, April 1990 and 1996

Location	Date	N	A.hu	A.cu	A.as	A.vi	A.pe	A.sy	A.ex	Und.
Louisiana	16 Apr. 1996– 3 May 1996	30				0/2		1/19		1/7
Georgia	* Apr. 1996	1				0/1
Florida
San Blas	* Jan. 1988	13						8/4		0/1
St. Marks	29 Oct. 1994	94		1		2/1	1	24/34	1/1	15/14
Lake Jem	2 Apr. 1995	16						12/3		0/1
Sanibel	20 Nov. 1995	24						2/2		14/6
Cross Creek	6 Apr. 1987	21	1	1/1	1	4/2		5		6
	15 Apr. 1988	2						1/1
	* Apr. 1990	17						4/10	2	0/1
	4–24 Apr. 1993	43				1		30/5		6/1
	3–22 Apr. 1994	56		0/1		5/2	1	27/7		7/6
Grand total		317	10.3%	41.3%	10.3%	206.0%	20.6%	199 63%	4 1.3%	86 27%

Most of the St. Marks monarchs collected in October (Table 2) had the A. syriaca fingerprint so they probably came from the Midwest where this Asclepias sp. is dominant (https://plants.usda.gov/). In contrast, the isotopic results showed that monarchs caught in St. Marks later in the migratory season (early December) likely originated from the southern (N = 5; Fig. 3, A1, A8, A10, A11, A19), south-central (N = 5; Fig. 3, A2, A5, A6, A14, A17), central (N = 8; Fig. 3, A3, A9, A12, A13, A15, A16, A18, A21) and northern regions (N = 2; Fig. 3, A7, A20). This suggests that St. Marks monarchs that arrive late in the overall autumn migratory period have different natal grounds from the ones collected early in their October migration (Table 2). Although the assignment to origin results indicate that some individuals captured in St. Marks may have originated in Cuba or Yucatán, we find this highly unlikely because individuals migrate south and not north during the autumn.

Figure 3.

Individual maps depicting probable origins of the twenty monarchs collected in St. Marks in December 2015. The blue area represents the most likely region of origin and the yellow area indicates areas with the least likely origins. Eastern monarch breeding grounds including Cuba and the Yucatán Peninsula, Mexico, are outlined in black. The monarch’s sex and deuterium value, δ²H, and first letter of the month that the monarch was collected are indicated. Only butterflies which had δ₂H values < -90‰ (VSMOW) and/or TLC fingerprints from North America were included, explaining the discrepancies in numbers of specimens between Table 1 and this figure.

Ten of the 21 monarchs captured in St. Marks in early December came from the south or south-central USA where A. viridis, A. humistrata, A. perennis and A. asperula capricornu—the last in south-central USA—are abundant and widely used by monarchs. These fingerprints were also present in Miami and Cuban monarchs (Tables 3–4) but absent at the Mexican overwintering colonies (Malcolm et al., 1993). Wing δ²H isotopic values of monarchs at the overwintering colonies in Mexico support different natal grounds for Florida-Cuba vs. Mexican overwintering monarchs. In the latter case, between 38% (Flockhart et al., 2017) and 50% come from the Midwest (Wassenaar & Hobson, 1998) where A. syriaca is abundant. These results suggest that most monarchs from the south and south-central USA may turn east, enter the Florida Peninsula, and continue to the insular and continental Caribbean. The presence of a monarch with the A. humistrata fingerprint in northern Venezuela (Fig. 4) suggests that a few may reach northern South America. Asclepias humistrata is native to eastern North America (Wild Flower Center.”Asclepias humistrata” (Online). University of Texas at Austin, TX, Aug. 27, 2015. Accessed 2 February 2023. https://www.wildflower.org/plants/result.php?id_plant=ashu3) and is absent in the Caribbean and South America (Woodson, 1954). However, this result has to be taken with caution since it is only one butterfly with a possible A. humistrata fingerprint, and the fingerprint is faded. A similar scenario was proposed by Urquhart (1987) where eastern North America monarchs will migrate through the insular Caribbean until they reach northern South America.

Figure 4.

TLC of monarchs raised on A. humistrata (A.hu), second and fourth channels from left, along with one monarch (A.hu-V) captured in an old pasture in the outskirts of Guanarito, Portuguesa State, Venezuela, 26 November 1995 (Supporting Information, Table S2). This is a doctored image composed of different channels assembled from two A. humistrata fingerprints channels and the digitoxin standard ‘DIG’ that came from an unpublished plate by Ron Martin and Steve Lynch and channel of a monarch captured in Venezuela.

Table 3.

TLC fingerprints of monarchs collected in the Miami area, south Florida, during 12 months starting in September 1994 and a sample from April 1990 (shown in italics). The fingerprints include A. humistrata (A.hu), A. curassavica (A.cu), A. asperula capricornu (A.as), A. viridis (A.vi), A. perennis (A.pe), A. syriaca (A.sy), and undetermined (Und.) fingerprints. The first number in each cell shows the number of females followed by males with that particular fingerprint and when there is a single number is/are a female/s. Supporting Information, Figure 4 shows the TLC fingerprints exhibited by these south Florida monarchs

Date	N	A.hu	A.cu	A.as	A.vi	A.pe	A. sy	Und.
1994
8 Sep.	29	9/19						0/1
22 Oct.	38	6/7	10/7					5/3
17 Dec.	45		14/4	10/12	1	1	1	1/1
1995
1 Jan.	45		23/20		1			1
10 Feb.	35		11/3			0/21
11, 12 Mar.	28		9/7		1	1/10
8 April	31		10/11		3/3			0/4
1990 14 April	44		3/6		10/19		0/1	3/2
1995 8, 9 May	36		3/5		9/19
20 July	10		3/1		3/3
10 Sep.	20	2	3		1/3			2/9
26 Oct.	10			1	4/5
6, 7 Nov.	40		1/1	1	2/10	1	12/5	3/4
Total	411	43 10.5%	155 37.7%	24 5.8%	97 23.6%	34 8.3%	19 4.6%	399.5%

Table 4.

Summary of TLC fingerprint results for monarchs collected in Cuba in Guanahacabibes, San Antonio and Zapata Swamp during 1993 and 1995–1997. The column labelled as ‘Und.’ accounts for undetermined fingerprints. The fingerprints included A. humistrata (A.hum), A. curassavica (A.cu), A. asperula capricornu (A.as), A. viridis (A.vi), A. perennis (A.pe) and A. syriaca (A.sy). The first number in each cell shows the number of females followed by males with that particular fingerprint and when there is only one number is/are a female/s. Supporting Information, Figure 4 shows the TLC fingerprints exhibited by these southern Florida monarchs

Site and date	N	A.hu	A.cu	A.as	A.vi	A.pe	A.sy	Und.
Guanahacabibes
Nov. 1993	15	0/1			0/2		1/8	1/2
Nov. 1995	3	0/1		0/1	1
San Antonio
Mar. 1995	10		2/6		1/0			0/1
Nov. 1995	18	1/1	9/4					3
Nov. 1996	53	1/2	4/8		12/11	0/2	1/2	2/8
Nov. 1997	38	2/2	8/7		1/2		2	8/6
Zapata Swamp
Nov. 1995	1						1
Total	138	11	48	1	30	2	15	31
		8%	35%	0.7%	22%	1.4%	11%	22%

Most monarchs captured in Miami had fingerprints from Asclepias species that grow in the south-eastern USA (Table 3; Supporting Information, Fig. S5). Monarchs with the same TLC fingerprints arrived in Miami in waves with little or no overlap with monarchs with other fingerprints, except those with the A. viridis fingerprint. Monarchs with the A. viridis fingerprint arrived during the spring, coinciding with this Asclepias species’ first and bigger peak of abundance (Tracy et al., 2022) and with Urquhart’s (1987) and Smith et al.,’s (1994) observations of high numbers of monarchs in Florida during the spring. Smith et al., (1994) observed that the monarchs appeared to come from the north. In early autumn, monarchs with the A. humistrata fingerprint were predominant, followed by a second smaller peak with both A. viridis and A. syriaca and, at the end of this season, A. asperula monarchs were dominant. In winter, February through to March, monarchs with A. perennis were predominant indicating a Type II migration behaviour where northern individuals pass through these regions before southern-breeding individuals. The most dominant fingerprint though the year in Miami was A. viridis (23.6%). We did not take into account the prevalence of the fingerprint of A. curassavica , as this is an introduced species and probably did not have a role in the evolution of the monarch migration. Asclepias curassavica is an exotic that probably originated in South or Central America (Woodson, 1954,) and is predominantly found in the south-eastern USA but is widely distributed in the Americas (Woodson, 1954; Satterfield et al., 2015).

In contrast, the monarch butterfly originated in the southern USA or northern Mexico (Zhan et al., 2014). Asclepias curassavica prevalence is explained by this species’ abundance throughout the year in the Miami pasture where monarchs were collected (Table 3). Monarchs with the A. syriaca fingerprint only represented 4.6% of the fingerprints in Miami (Table 3), in stark contrast with 92% at the Mexican overwintering colonies (Malcolm et al., 1993).

Cuba

During November, Cuba received an influx of migrants with similar TLC fingerprints as Miami (Tables 3–4), but monarchs from the two collecting locations in Cuba, Guanahacabibes and San Antonio, differed in their TLC fingerprints, sex ratios and phenotype. Guanahacabibes, the most western portion of Cuba (Fig. 2), had predominantly males with A. syriaca fingerprints followed by A. viridis, and a few individuals showed A. humistrata and A. asperula fingerprints. On the other hand, San Antonio had roughly the same amounts of males and females with mostly A. viridis and few A. syriaca fingerprints (Table 4) (without taking in account A. curassavica).

The assignment to origin analysis using stable isotopes of the Cuba monarchs show that they likely have natal origins from eastern North America (Figs 5–7). Some of the assignment to origin results indicate that some butterflies may have originated in Cuba or Yucatán, especially for individuals with δ²H > -100‰; however, a good number of these butterflies have fingerprints of Asclepias spp. that only grow in North America indicating that they probably originated there. The monarchs that do not have the TLC fingerprints but have the Cuba and the Yucatán assignment of origin, could support a regional movement of monarchs around the Caribbean, Mexico and Guatemala as suggested by the presence of the Puerto Rican subspecies, D. p. portoricensis, in these areas (Supporting Information, Figs S10, S11, S12).

Figure 5.

Individual maps depicting probable origins of 14 monarchs collected in November 1993 and 1995 in Guanahacabibes, Cuba, most western portion (Fig. 2). The blue area represents the most likely region of origin and the yellow area indicate areas with the least likely origins. Eastern monarch breeding grounds including Cuba and the Yucatán Peninsula, Mexico, are outlined in black. When available, the monarch’s sex, deuterium value, δ²H, first letter of the month that the monarch was collected, and TLC Asclepias fingerprint are displayed in the individual maps. See Table 2 for species names associated with abbreviations. Only butterflies which had δ₂H values < -90‰ (VSMOW) and/or TLC fingerprints from North America were included, explaining the discrepancies in numbers of specimens between Table 4 and this figure.

Figure 6.

Depictions of probable origins for individual monarchs collected in November 1995 and 1996 in San Antonio, Cuba (Table 4). The blue area represents the most likely region of origin and the yellow area indicate areas with the least likely origins. Eastern monarch breeding grounds including Cuba and the Yucatán Peninsula, Mexico, are outlined in black. When available, the monarch’s sex, deuterium value, δ²H, first letter of the month that the monarch was collected, and TLC Asclepias fingerprint are displayed in the individual maps. See Table 2 for species names associated with abbreviations. Female C3 (Supporting Information, Fig. S6) was the only individual collected in March 1995. Only butterflies which had δ₂H values < -90‰ (VSMOW) and/or TLC fingerprints from North America were included, explaining the discrepancies in numbers of specimens between Table 4 and this figure.

Figure 7.

Individual depictions of probable origins of monarchs collected in November 1997 in San Antonio, Cuba (Table 4). The blue area represents the most likely region of origin and the yellow area indicate areas with the least likely origins. Eastern monarch breeding grounds including Cuba and the Yucatán Peninsula, Mexico, are outlined in black. When available, the monarch’s sex, deuterium value, δ²H, first letter of the month that the monarch was collected, and TLC Asclepias fingerprint are displayed in the individual maps. See Table 2 for species names associated with abbreviations. Only butterflies which had δ₂H values < -90‰ (VSMOW) and/or TLC fingerprints from North America were included, explaining the discrepancies in numbers of specimens between Table 4 and this figure.

Most monarchs collected in Guanahacabibes had the fingerprint of the abundant Midwest A. syriaca; however, four of nine monarchs with this fingerprint came from the south-central (N = 4; Fig. 5, B3, B5, B9, B11) or northern distribution range of this plant (N = 2; Fig. 5, B4, B8), and the remaining butterflies came from the south or south-central USA and had an A. humistrata or A. viridis fingerprint (N = 3; Fig. 5, B1, B6, B7).

The natal grounds of monarchs collected from San Antonio in central Cuba (Fig. 2) in November 1995–1997, show that the majority of monarchs collected here came from the south or south-central USA where A. viridis, A. humistrata, A. perennis, A. curassavica and A. asperula capricornu—the last one in the south-west and south-central USA—are abundant (Supporting Information, Fig. S5; Figs 6–7). Monarchs collected in San Antonio with the A. syriaca fingerprint originated mostly from this Asclepias species’ southern range boundary (N = 2; Fig. 6, C82, C86) or northern range (Fig. 7, D29). There was also a monarch with an A. curassavica fingerprint that originated from the north (Fig. 7, D33). Asclepias curassavica is an introduced species but is widely planted (CABI, 2022).

An important number of monarchs captured in December in St. Marks (Fig. 2) and most of the monarchs captured in Miami and San Antonio, Cuba, differ in their natal grounds from the overwintering monarchs in Mexico, but these two groups of monarchs also differ from Mexican monarchs in two key migratory traits: wing condition and fat mass. Miami and San Antonio monarchs had wings in worse condition and had appreciably less fat than did Mexican overwintering monarchs (Supporting Information, Tables S2–S3). On the other hand, St. Marks monarchs collected in October with the A. syriaca fingerprint had a lower fat content compared to October migrants in Texas and the Mexican overwintering monarchs collected in November (Supporting Information, Table S2; Dockx, 2012).

In Cuba, a monarch collected in March with the A. viridis fingerprint that probably came from the south-eastern USA where this Asclepias sp. is present, had a wing condition of 3 (Fig. 6; Supporting Information, Fig. S6), suggesting that this individual monarch was staying on the island or was returning to the USA possibly from the continental or insular Caribbean. Most monarchs returning from the Mexican overwintering colonies had a wing condition between 3 and 3.5 (Malcolm et al., 1993).

Monarchs that displayed A. syriaca and A. viridis and that were collected in Guanahacabibes and San Antonio differ in their phenotype. Monarchs in Guanahacabibes with these fingerprints were larger and had wings in better condition than those from San Antonio (Supporting Information, Table S3). Although A. syriaca monarchs in Guanahacabibes and the monarchs at the Mexican overwintering colonies had similar wing size, the Guanahacabibes monarch wings were in worse condition, and had less lipid and lean mass content than their Mexican conspecific (Dockx, 2012; Supporting Information, Tables S2–S3). Furthermore, the majority of the Miami (96%) and Cuban (84%) monarchs had mated, whereas the Mexican overwintering monarchs were in reproductive diapause.

Yucatán

In Yucatán, Mexico, west of Guanahacabibes, assignments to origin analysis using stable isotopes showed that monarchs collected there had similar natal grounds to monarchs from Guanahacabibes (Figs 5 and 9), a combination of south-east, east-central and north-east USA but also some potentially originated in Cuba or the Yucatán, particularly individuals with wing δ₂H isotope values > -100‰. Monarchs that had Cuba or the Yucatán assignment of origin could support a regional movement of monarchs suggested by the presence of the Puerto Rican subspecies, D. p. portoricensis, in Cuba, Mexico and Guatemala (Supporting Information, Figs 10–12).

Most Yucatán monarchs were collected from late September through November coinciding with the movement of North American monarchs to the Caribbean. However, two monarchs that were collected in this Mexican peninsula in March and June (N = 2; Fig. 8, E12 and E18, respectively) probably originated from the south and/or south-central USA, Yucatán or Cuba. The monarch collected in March had the D. p. plexippus phenotype and intermediate wing condition (Supporting Information, Fig. S7; Fig. 8, E12; Fig. 9) suggesting that these monarchs were spending the winter in Yucatán, as monarchs do in Florida, or they were returning to their natal grounds in North America. The female, collected in Yucatán in June (Supporting Information, Fig. S8; Fig. 8, E18; Fig. 9), had a wing condition of 3 and a similar wing shape to the monarch collected in Cuba in March with the A. viridis fingerprint (Supporting Information, Fig. S6). A wave of monarchs with the A. viridis fingerprint, that was present in Miami in May had an average wing condition of 2.4 (Supporting Information, Table S2). Furthermore, we know that most of the monarchs that arrive in Miami in May leave and only a few remain there in the summer (Table 3).

Figure 8.

Depictions of probable origins of individual monarchs collected in the Yucatán. The blue area represents the most likely region of origin and the yellow area indicate areas with the least likely origins. Eastern monarch breeding grounds including Cuba and the Yucatán Peninsula, Mexico, are outlined in black. The sex, its individual deuterium value, δ²H, and first letter of the month that the monarch was collected are shown. See Table 2 for species names associated with abbreviations. Only butterflies which had δ₂H values < -90‰ (VSMOW) and/or TLC fingerprints from North America were included, explaining the discrepancies in numbers of specimens between Table 1 and this figure.

Figure 9.

Locations where monarchs were collected in Veracruz, the Yucatán Peninsula and Guatemala plus the location where the egg and larvae census (Supporting Information, Table S4) was conducted. The letter and the number are the same used in the assignment of origin depictions of the isotope analyses in Figures 8 and 10. The locations marked with a cross, triangle or a diamond are where Welling collected (Table 5), squares represent individuals from the ECOSUR collection (Fig. 8) and circles represent the arthropod collection from Universidad del Valle de Guatemala (Fig. 10).

Guatemala

Guatemalan monarchs, collected in June through November in 1965 and 1966 (Newcomer, 1967) had the A. viridis fingerprint and monarchs collected in September and October had the A. asperula capricornu fingerprint (Table 5). Our data shows the presence of these two fingerprints in Miami, Cross Creek, Florida, and in Guanahacabibes, Cuba (Tables 2–4). Even though A. asperula capricornu is very abundant in the south-western and south-central USA, Malcolm et al., (1993) did not find any returning migrants in the southern (early spring) or northern (late spring) USA with this Asclepias species fingerprint. Asclepias viridis and A. asperula capricornu are native to North America, A. asperula capricornu extends into northern Mexico and A. viridis is absent from Mexico and Guatemala (Juárez-Jaimes et al., 2007). Our data suggest that monarchs with the A. asperula and A. viridis fingerprints enter Florida, continue to Cuba and may then move to Guatemala.

Table 5.

TLC fingerprints of monarchs collected by Welling in two locations in Guatemala in 1965 and 1966 and Mexico in September 1964 and 1975 (Fig. 9). The following abbreviations were used for the localities: Chimalte (Department of Chimaltenango); Alta Ver. (Department of Alta Verapaz) (Newcomer, 1967)

Country	Location	Date	A.vi	A.sy	A.as
Guatema	Chimalte	26 Sept. 1965	1
		28 Nov.1965	1
	Alta Ver.	19 Jun. 1966	1
		26 Jun. 1966	1
		2 July 1966	1
		11 July 1966	1
		4 Aug. 1966	1
		16 Aug. 1966	1
		1 Sept. 1966	1
		10 Sept. 1966	2
		20 Sept. 1966	2
		20 Sept.1966			2
		27 Sept. 1966	2		1
		4 Oct. 1966	1
		7 Oct. 1966
Mexico	Catemaco	20 Sept. 1964	1§		1
		20 Sept. 1964
		20 Sept. 1964		1*
	Colima	15 Sept. 1975	1ǂ	2

*Faint fingerprint of possible A. syriaca.

§Supporting Information (Fig. S13).

ǂSupporting Information (Fig. S14).

The isotopic results for Guatemalan monarchs from Universidad del Valle de Guatemala were consistent with origins from the north to the south in eastern North America (Fig. 10), as was observed in Guanahacabibes (Fig. 5) and Yucatán (Fig. 8), suggesting that they are part of a common migratory route. One female was caught in the autumn (Fig. 9; Fig. 10, F7) in the Sierra de las Minas Biosphere Reserve, where peaks ≥ 3000 m a.s.l. are found and where Abies guatemalensis, a counterpart to the Mexican fir Abies religiosa that grows at the monarch overwintering colonies in Mexico, occurs. Two individuals, caught in the winter in Guatemala, came from north-central and central North America (N = 2; Fig. 9; Fig. 10, F1, F6) and one captured in the spring, came from central North America (Fig. 10, F11). These monarchs, collected in winter and spring, were not recently eclosed since their wing conditions ranged from 3.5 to 4.5. In contrast, two monarchs, collected in July from central and south-eastern North America had intermediate wing conditions, 2 and 2.5, (Supporting Information, Fig. S9; Fig. 9; Fig. 10, F3, F4) that suggests their recent arrival. These two monarchs captured in Guatemala in July have similar wing condition and natal grounds as the monarch collected in the Yucatán in June (Supporting Information, Fig. S8; Fig. 8, E18; Fig. 9), suggesting that they are part of the same group of migrant monarchs.

Figure 10.

Depictions of origins of individual monarchs from the arthropod collection of Universidad del Valle in Guatemala (Fig. 9). The blue area represents the most likely region of origin and the yellow area indicate areas with the least likely origins. Eastern monarch breeding grounds including Cuba and the Yucatán Peninsula, Mexico, are outlined in black. The monarch’s sex, deuterium value, δ²H, and first letter(s) of the month that the monarch was collected are provided. Only butterflies which had δ₂H values < -90‰ (VSMOW) and/or TLC fingerprints from North America were included, explaining the discrepancies in numbers of specimens between Table 1 and this figure.

DNA results

The samples of Aruba, Ecuador, Puerto Rico and Cuba (this last one with one exception, D29 in Figs 11–12), each form an independent and distinct lineage (Fig. 11) but derived from the North America population (Fig. 12); however, the absence of population structure between North and Central America and Bermuda suggest gene flow between these populations (Fig. 11).

Figure 11.

PCA plot of the genomes of the 82 monarchs based on the first two principal components. The probable natal grounds of the first outlier, D29, is shown in Figure 7 and for the second one, F10 (Supporting Information, Fig. S9). The abbreviations are the same as those used in Figure 12.

Figure 12.

Neighbour-joining consensus tree of 82 individuals collected in the Americas from Zhan et al., (2014) plus an additional 21 monarchs from St. Marks, Florida (collected late in their migratory season), Cuba and Guatemala. The colours represent the different localities where the monarchs were collected: STM, St. Marks from December; NAM-MEX, Mexico; NAM-CA, California; STM-2014, St. Marks from October, 2014; NAM-NJ, New Jersey; NAM-MA, Massachusetts; SFL, South Florida; GUA, Guatemala; BLZ, Belize; NAM-TX Texas; BMU, Bermuda; CUB, Cuba; CRC, Costa Rica; ABW, Aruba; ECU, Ecuador; PRI, Puerto Rico. The letter and the number that follow some of the specimens refer to the isotopic natal ground in Figures 3, 5–8 and 10. In addition, a maximum likelihood method tree is included to explore the phylogenetic relations between individuals (Supporting Information, Fig. S15).

Six of the seven monarchs collected in South Florida are clustered between the mainly North American and the rest of the Americas’ monarchs; the seventh is the only south Florida monarch collected during the winter that is nested within the North American monarchs and shares a very similar genetic signature with the St. Marks monarch collected in December (Fig. 12).

Two-thirds of the consensus tree (Fig. 12) grouped mostly monarchs collected in North America with the exception of two monarchs, one collected in Guatemala and the second in Belize, showing that these two migrant monarchs probably had no genetic exchange with the resident populations in Central America unlike other monarchs collected in Belize and Guatemala (Fig. 13). Monarchs moved out of North America through Central America and the Florida Peninsula and moved from South Florida to Cuba, where we analysed six monarchs for their DNA. All the Cuba monarchs were captured in the central part of the island, San Antonio; five showed an almost even mixture of North American vs. Caribbean, Central and South America monarch populations (N = 5; Fig. 13, D4, D10, D20, D33, D42); three were reproductively active females. Two of these three were migrants (Fig. 7, D10, D29); the third was a Cuban resident (Figs 12–13, D42). Cuba resident D42 and migrant D10 had eggs. All these observations suggest that some migrants that arrive in Cuba interbreed with the resident population and thus leave the North American genetic signature in this Caribbean island. Then, when they return to the USA, they bring the genetic resident population signature in their fertilized eggs, which results in the almost 50:50 ratio mixed ancestry. In contrast to the monarchs with an almost even admixture, migrant female D29 (Figs 7, 13) had a ~90% North American signature. This female came from the monarch northern breeding boundary, had the A. syriaca fingerprint (Fig. 7, D29) and was genetically closest to two monarchs that originated in the Atlantic region (Massachusetts, USA and Bermuda) (Fig. 12). Brindza et al., (2008) found that Atlantic coastal monarchs have a significantly lower chance of arriving at the overwintering colonies in Mexico.

Figure 13.

Genetic structure, ancestry and estimated admixture proportions in each individual genome and group of monarchs of different sizes (Zhan et al., 2014) plus Cuba, Guatemala, and an additional 21 monarchs from St. Marks that were collected late in their migratory season. The colours represent admixture proportions in each individual genome. South FL, South Florida. ‘K’ represents the predicted number of ancestral clusters used in the model.

In Puerto Rico, the genetic presence of North American monarchs was not detected, and it appears that this group of monarchs is derived from Cuba (Fig. 13). Puerto Rico has the distinctive subspecies D. p. portoricensis, a subspecies also found in other Caribbean areas including Cuba, Mexico and Guatemala (Supporting Information, Figs S10–S12). A North American male (Fig. 7, D21), with the typical D. p. portoricensis phenotype (Supporting Information, Fig. S10), was caught in San Antonio-Cuba which suggests that the monarch genes moved in both north and south directions. Monarchs in Aruba are genetically closer to monarchs in Ecuador, which supports the arrival of northern monarchs to South America through the Caribbean.

In Central America, the North American genetic presence is more prevalent than in the Caribbean, as was found in Belize where it was around 87%, and where one individual had an almost 100% North American signature. In Guatemala the North America signature drops to 78% but, as in Belize, there was one monarch, F10, with almost 100% North American DNA. Female F10 was not included in Figure 10 because its δ²H was -88‰ (VSMOW) but its natal ground is presented in the Supporting Information (Fig. S9). Further south, in Costa Rica, the North American monarch DNA signature decreases but is still prevalent (76%), and reaches northern South America with 44% in Ecuador.

From North America, monarchs appeared to take two routes: one through Central America and the other through south Florida into the continental and insular Caribbean. Monarchs appeared to arrive in South America through the Lesser Antilles and Central America (Figs 11–12).

DISCUSSION

In addition to the well-known migration of the eastern population of monarch butterflies to Mexico, eastern North American monarchs also migrate to Cuba, Yucatán and Guatemala. We provide evidence using multiple intrinsic markers and phenotypic traits that these latter individuals are part of different, albeit more minor, migratory routes, one that passes through the south-eastern USA that we call the ‘south-eastern route’ (route 1, Fig. 14) and the Atlantic migratory route (route 2, Fig. 14). Monarchs from these migratory routes have defined natal grounds, feed on particular Asclepias species as larvae, have their own migratory routes and times, as well as specific phenotypes.

Figure 14.

Proposed migratory routes 1 and 2. Direct lines of evidence are presented for routes or portions of it by solid lines, circumstantial lines of evidence are showed by broken ones.

We also have some indication of how the monarchs return to the USA. These routes may have been previously overlooked because they were considered ‘aberrant’ (Urquhart, 1987) or dispersal routes (Brower, 1995), and monarchs that follow them do not fly, at least currently, in large numbers, their migration lasts longer than that of migratory monarchs that winter in Mexico, and the possible overwintering locations in Guatemala have not been discovered. Moreover, monarchs using these migratory routes may have declined with increased human densities over the past century further obscuring their prevalence.

The south-eastern migration route (route 1) involves monarchs that mostly originate from the south-east USA where A. viridis, A. humistrata, A. perennis and A. asperula capricornu (the latter in south-central USA) are dominant. These migrant monarchs will move mostly in a south-east direction, enter the Florida Peninsula, then move to Cuba, continue to the Yucatán Peninsula, and some will continue south-west to overwinter in Guatemala. Two monarchs tagged in Guatemala and later recaptured in Chiapas, Mexico (J. Schuster, pers. comm.) suggest that some of these monarchs return to the USA breeding grounds. Monarchs not only return to the USA via Mexico but also our data and anecdotal observations suggest that some monarchs from this south-east and Atlantic migratory route return to the USA through Florida during the spring as well.

Florida

The timing of the southward migration of these south-east monarchs showed close synchrony with their host plant phenology, mostly A. viridis, A. asperula, A. humistrata and A. perennis. This migration lasted more than 6 months, and waves of monarchs, mostly from the south-eastern USA, arrived in Miami with particular TLC fingerprints and/or natal grounds. We only detected these waves with particular TLC fingerprints in Miami, because here, they were sampled almost every month for 12 months. This contrasts with Cuba where we sampled monarchs in November for 3 years and once in March. There was little or no overlap in the waves of monarchs with particular fingerprints. The first wave, in April and May, consisted of monarchs that fed on A. viridis and the last wave, in December, was of monarchs that fed on A. asperula. Asclepias perennis was the most common fingerprint for monarchs captured in Miami in the winter, with 21 captured in February and 11 captured in March (Table 3), all but one of which were males. In the north Florida Big Bend region, females were observed laying eggs on the abundant A. perennis, an aquatic milkweed (S.A. Davis, pers. obs.; Kirschke et al., in prep.). Males appear to be dominant in north Florida during the winter but egg counts indicate that there are enough females laying eggs to keep the generations going (T. Van Hook, pers. comm.). Davis (pers. obs.) and Kirschke et al. (in prep.) observed adults and larvae at different stages during winter and early spring in north and central Florida. Unlike all other native milkweeds, A. perennis remains green during the winter in a big portion of the south-eastern USA. Furthermore, the Big Bend region has very high densities of native milkweed species that serve year-round as food plants for monarchs (S.A. Davis, pers. obs.).

Wing condition of the A. perennis butterflies in Miami showed that they were a combination of newly-hatched and old butterflies. The presence of freshly eclosed butterflies supports S.A. Davis and Kirschke et al.’s observations of numerous monarch larvae feeding on A. perennis in the Big Bend coastal area from September to December, March and April. Since Miami is not part of the Mexican return migratory route, the most probable explanation for monarchs with intermediate and old wing conditions during winter and early spring is that they overwintered in this southern part of the Florida Peninsula and/or they were returning from Cuba or other areas of the insular and continental Caribbean. The A. perennis fingerprint occurred in monarchs collected in October, at St. Marks in the Florida Panhandle (Fig. 2; Table 2) and in November in Miami (Table 3) and Cuba (Table 4); monarchs with this Asclepias sp. fingerprint have very versatile behaviours: some spend the winter in Florida, others reproduce there and others continue to Cuba.

Monarchs with the A. viridis fingerprint were present in almost every month of the year in Miami. The two picks of abundance of a A. viridis (Tracy et al., 2022) were matched by higher numbers of monarchs in southern Florida (Smith et al., 1994). The October monarch peak corresponds to autumn migration through Florida and we propose that the second peak in April and May is the result of some monarchs feeding on A. viridis in the spring, and then turn east into the Florida Peninsula and Cuba, instead of continuing north and recolonizing the USA. Lynch & Martin (1987,) caught 60 monarchs with the A. viridis fingerprint between late March and early April in Louisiana and in Gainesville, northern Florida, Malcolm collected 18 monarchs with the A. viridis and 22 with the A. humistrata fingerprint during the spring (Malcolm et al., 1993).

The A. viridis peak in April and May is larger than the one in October (Tracy et al., 2022) as well as the peak for the monarchs in Miami (Table 3). Asclepias viridis and A. asperula are the most abundant and widespread Asclepias spp. encountered by the spring-migrating monarchs in their flight through Texas and Louisiana (Lynch & Martin, 1987). However, Malcolm et al. (1993) found in the spring that only 4% of monarchs in the south-western USA and none in the south-east had the A. viridis fingerprint. This supports the conclusion that an important number of monarchs with the A. viridis fingerprint in the south-east do not move north. In contrast, the A. viridis fingerprint was the most abundant group in the spring in south Florida, (Tables 2–3). Similarly, A. asperula is another of the most abundant plants that the re-migrant monarchs encounter in the south-eastern USA, nevertheless, Malcolm et al., (1993) found no monarchs with the A. asperula fingerprint during the spring. We suggest that most of the monarchs with this Asclepias sp. fingerprint follow one of two routes, first through Mexico or second, through Florida and Cuba (south-eastern route in Fig. 14; Tables 3–4) until reaching Guatemala (Table 5). Support for the arrival of monarchs with the A. asperula and A. viridis fingerprint into Miami in December and in spring comes from Zanden Vander et al.’s work (2018). They found that the Texas-Oklahoma border, where both of these Asclepias spp. are dominant, represents one of the most prominent natal grounds for monarchs collected in south Florida during the winter and spring.

Florida to Cuba

Monarchs with A. viridis or A. asperula fingerprints were present in Cuba and Miami suggesting that they are part of the same migratory route. However, Cuba appears to be where eastern North America monarchs may converge and then split into different routes: east, central or west. The eastern and south-eastern movement of monarchs in Cuba and through the insular Caribbean is supported by recoveries of tagged monarchs in Jamaica, Puerto Rico and Trinidad (Urquhart, 1987) and reports of roosting monarchs in the mountains of Punta Maisi, in the most eastern portion of Cuba (L. R. Hernandez, pers. comm.). This eastern movement of monarchs in Cuba is part of route 2 (Fig. 14).

San Antonio, in central Cuba, received migrant monarchs every November (Table 4) during 3 years and one time in March that mainly fed on the Asclepias spp. that are dominant in the south-eastern USA: A. viridis, A. humistrata and A. perennis, and some A. syriaca (mostly a Midwest Asclepias sp.). In Cuba, migrants were always present when monarchs were sampled for 3 years in November and one time in March. We suspect that North American monarchs are on the island in other months. Only 2.5% of the monarchs collected in San Antonio in November had the A. syriaca fingerprint and came from this Asclepias species’ northern and southern distribution boundaries. In contrast, the majority of Mexican overwintering monarchs had the A. syriaca fingerprint (Malcolm et al., 1993) and came mostly from the Midwestern USA and surrounding areas (Wassenaar & Hobson, 1998).

TLC and/or isotopes of monarchs collected in Guanahacabibes in November indicated that most were migrants that had fed on A. syriaca, the dominant host plant identified in the Mexican overwintering colonies (Malcolm et al., 1993). Monarchs collected in Guanahacabibes with this A. syriaca fingerprint came from the entire distribution range of this Asclepias species (Fig. 5; Supporting Information, Fig. S5).

Guanahacabibes, western Cuba to Yucatán

From Guanahacabibes, monarchs continue to Yucatán, as shown by Urquhart’s (1987) recapture of four tagged monarchs in the Yucatán. His observations in the Yucatán, and Dockx’s observations in Guanahacabibes in November suggest that monarchs will fly to this Peninsula from Guanahacabibes at the end of the year. Dockx spent several days at the western tip of San Antonio Peninsula in Guanahacabibes (Fig. 2) in November for 2 years in the late 1990s and observed that monarchs waited until the winds were blowing in a westerly direction towards the Yucatán to fly in that direction. Urquhart (1987) reported four Canadian tagged monarchs that were later recaptured in the Yucatán Peninsula (no date available). There is evidence of thousands of monarchs roosting at the end of the year in different locations in the Yucatán from reports by Barbara MacKinnon, who worked with Urquhart in the Yucatán, G. Cobb’s observations and a YouTube video (https://www.youtube.com/user/Yucamama). De la Maza (1995) and Chávez (1995) reported significant numbers of monarchs from October through March in the Yucatán. Furthermore, the isotopic results of the Yucatán monarchs showed the arrival of migrant monarchs with similar natal grounds to Guanahacabibes, a combination of south-eastern, east-central and north-eastern USA. North American monarchs were present in Yucatán from September through November, and March and coincide with the presence of migrants in Miami from September to February and in November and March in Cuba.

FLORIDAN AND CUBAN PHENOTYPIC TRAITS

Monarchs on the south-east route not only share natal grounds but also migratory routes, destinations and phenotypic traits that are important in their migration: reproductive status, wing size and condition, lean and fat mass. The Mexican migrants are mostly in reproductive diapause, have wings in better condition, and have higher fat and lean mass content (body mass devoid of fat). This enables the latter migrants to have a faster migration, arrive on time, overwinter successfully and undertake spring remigration. In contrast, the majority of Floridan-Cuban migrants were reproductively active, had wings in poorer conditions, and had lower fat and lean mass content (Supporting Information, Tables S2–S3; Dockx, 2012), suggesting a more opportunistic and flexible migration strategy as Floridan-Cuban monarchs encounter year round resources in the Neotropical and subtropical regions compared to the Mexican monarchs in temperate areas. This opportunistic and flexible behaviour is revealed in the wide behaviours that Florida-Cuban migrants exhibit: some monarchs, like the Cape San Blas (Table 2) and Honeymoon Island monarchs (Knight, 1998), do not leave and instead spend the winter in the USA; others migrate to the insular and continental Caribbean; a group of monarchs probably overwinter in Guatemala; and their reproductive status suggests that several of them reproduce along the way.

Putative eastern North American monarchs are present in the Yucatán Peninsula from September through March, the same months that monarchs migrate to Mexico, overwinter, and/or start re-migrating. The wing condition found in these butterflies suggests that they spent their winter in the Yucatán. Also, the presence of a female monarch that was in Yucatán in June and had a wing condition of 3.5, and two monarchs in July in Guatemala, supports the movement of migrants through the Florida Peninsula at two different times, one that more or less coincides with the Mexican migration and the other between April and May (Table 3; Urquhart, 1987; Smith et al., 1994).

YucatÁn to Guatemala

We know from MacKinnon’s and Cobb’s observations and the video (https://www.youtube.com/user/Yucamama) that the thousands of roosting monarchs in the Yucatán Peninsula dispersed after a few days (B. MacKinnon, pers. comm.). MacKinnon and Cobb observed that when they departed, they flew in a southerly direction along the east coast. Most of the North American monarchs in Guatemala were caught in the Sierra Madre or in nearby mountainous regions, supporting Urquhart’s hypothesis that the monarchs overwinter in these areas (Urquhart, 1987). De la Maza (1995) reported that Mexican overwintering monarchs engage in downhill movements searching for water and nectar, then, after a few hours, return to their overwintering colonies. What happens to these North American monarchs after they arrive in Guatemala is unknown. All Guatemalan samples came from two collections with very limited field information. However, a count of eggs, larvae and pupae in the Sierra Madre in Guatemala (Supporting Information, Table S4) and Michael W. Dix’s observation of thousands of monarchs moving downslope in the Sierra de las Minas in Guatemala in late February in the 1980s (M. & M. Dix, pers. comm.) shows that some migrant monarchs overwinter and probably reproduce there. The recapture of two Guatemala-tagged monarchs in Chiapas, Mexico (J. Schuster, pers. comm.) shows that Guatemala monarchs can move north, and suggests how monarchs from the south-east migratory route may recolonize the USA.

Eggs and larvae were counted in Guatemala where a small breeding population of monarchs lived throughout the year in a pasture where A. curassavica was abundant, a similar scenario to the collecting sites in Miami and San Antonio, Cuba. Asclepias curassavica is native to South America (Woodson, 1954) but is abundant in tropical and subtropical areas worldwide (https://www.missouribotanicalgarden.org/ Missouri Botanical Garden. “Asclepias curassavica” (Online). St. Louis, MI. https://www.missouribotanicalgarden.org/plan-your-visit/family-of-attractions/butterfly-house/butterflies-and-plants/our-plant-collection/article/1522. Accessed 2 February 2023.). There was a large increase in eggs from September, peaking from October or November until the first months of the year, the same months that migrant monarchs were present in Miami, Cuba, the Yucatán and Guatemala (Supporting Information, Table S4; Tables 3–5; Figs 5–10). This increase in eggs may indicate the arrival of North American migrant monarchs and not be attributable to the winter or the rainy season (as this begins in May, peaks in August and September and ends in October; the dry season runs from November to April). Monarchs in Cuba, the Yucatán and Guatemala migrate at similar times, share natal grounds (Figs 5–8, 10), and share common DNA signatures (Fig. 13). These observations support our hypothesis that this group of monarchs is part of an extant migratory route. We do not have DNA ancestry data for the Mexican Yucatán; however, Zhan et al., (2014) sampled three monarchs from Belize that borders the Yucatán Peninsula and showed that Belize monarchs share a close ancestry with Guatemalan and North American monarchs (Fig. 13).

All the Cuban samples analysed for DNA ancestry came from the central area of the island and none was from the most western portion, Guanahacabibes, which we would expect to share a closer ancestry with the Belize and Guatemala monarchs. Most of the continental USA and Mexico samples show unique ancestries with almost no mixing; in contrast, Cuba, Belize, Costa Rica, Guatemala and Bermuda exhibit a more diverse genetic ancestry (Fig. 13) which supports an input of different monarch populations to these areas.

The Atlantic migration (route 2) (Fig. 14) is comprised of monarchs that mainly migrate east of the Appalachians, enter the Florida Peninsula, fly to Cuba, and then go east into the Caribbean, with some of them perhaps reaching northern South America like the monarch with the A. humistrata fingerprint collected in northern Venezuela (Figs 1, 4; Supporting Information, Table S2).

Monarchs with the A. humistrata fingerprint move mostly east of the Appalachians (Malcolm et al., 1993), and Brindza et al. (2008) observed that the Atlantic coast migrant monarchs mostly do not end up in Mexico. This was supported by Steffy’s results (2015) that suggest that inland monarchs have a higher chance of reaching Mexico than monarchs that migrate through the Atlantic coast. The recaptures of the Monarch Monitoring Project, that has been tagging monarchs for over two decades along the Atlantic coast, especially in Cape May, New Jersey, show that monarchs move along the Atlantic coast (N = 35), Florida (N = 15), and some end in the Bahamas (N = 2); in contrast, only three monarchs were recovered in Mexico [Monarch Monitoring Project | New Jersey Audubon (njaudubon.org)]. In addition, these coastal monarchs had slightly but markedly smaller wings and lower wet mass than inland monarchs (Brindza et al., 2008). Furthermore, A. humistrata fingerprint monarchs collected in Miami generally had smaller wings, were in worse condition, and had less lean and fat mass than the A. syriaca monarchs at the Mexican overwintering colonies (Supporting Information, Table S2).

In Miami, during the autumn, monarchs with the A. humistrata fingerprint were predominant whereas the A. perennis fingerprint’s prevailed during the winter (Table 3). Asclepias perennis is an important host plant for North Florida monarchs (S. Davis, pers. comm. S. Davis). The monarchs with the A. perennis fingerprint in Miami had different ages as indicated by their wing conditions. Since Miami is not part of the Mexican re-migrant route, the most probable scenario for monarchs with intermediate or old wing conditions during the winter is that they had overwintered in the southern part of the Florida Peninsula and/or they were returning from Cuba (and perhaps other areas of the Caribbean?). Observations that monarchs in the Turks and Caicos fly north-west towards Florida support the proposal that monarchs fly towards the Florida Peninsula at the end of winter and beginning of spring. Moreover, monarchs were seen flying north in different areas of Florida’s east coast (Brown, pers. comm. C. Brown). Many monarchs were also present in the autumn in the Turks and Caicos Islands [Times of the islands. “The Butterflies of The Turks and Caicos Islands” (Online). Spring, 2005 (https://www.timespub.tc/2005/04/the-butterflies-of-the-turks-and-caicos-islands/) Accessed 2 February 2023]; Fig. 1; Supporting Information, Table S1]. Furthermore, 351 of the 386 monarchs collected at the Mexican overwintering sites had the A. syriaca fingerprint, and none had the A. humistrata or A. perennis fingerprint (Malcolm et al., 1993). In contrast, monarchs with these two fingerprints were among the most abundant fingerprints in Miami during the autumn and winter, without taking into account monarchs with A. curassavica.

Remigration through FloridaNot all monarchs leave the Florida Peninsula, some winter there and when spring arrives, they reproduce and move north. There are reports of winter-breeding monarchs in southern Florida, from the Everglades to Lake Okeechobee (Brower, 1961; Urquhart & Urquhart, 1976; Brower, 1985; Cohen, 1985). In the Florida Panhandle, monarchs overwintered in Cape San Blas (Table 2; Fig. 2) and Honeymoon Island (Knight, 1998,). These monarchs, collected in Cape San Blas in January and in February in Honeymoon Island, had the A. syriaca fingerprint and significantly lower lipid content than the Mexican overwintering monarchs (Knight, 1998; Dockx, 2012). This suggests that these monarchs had insufficient lipid reserves to survive overwintering in Mexico and that instead, they were spending the winter in Cape San Blas and Honeymoon Island. All eight females collected in Cape San Blas were mated. In central Florida, Honeymoon Island (Knight, 1998) and Lake Jem (Table 2), there were migrant monarchs with mostly the A. syriaca fingerprint during the winter and early Spring. Furthermore, there are numerous reports of monarchs spending the winter in several areas of Florida (https://monarchwatch.org/tagging).

After overwintering in several areas of Florida, monarchs appear to reproduce and move north. Van Hook observed eight monarchs in central Florida feeding and flying leisurely along the bank of the Oklawaha River at the edge of the Ocala National Forest on 8 April 1989 (T. Van Hook, pers. comm. in Dockx, 2012). These monarchs were all noticeably small and in excellent condition, which suggests that they were recently eclosed and probably the descendants of migrant monarchs that overwintered in Florida. In north-central Florida, in Cross Creek (Fig. 2), monarchs arrived and started oviposition on the emergent A. humistrata in late March and were gone by early May (Knight, 1998). In Georgia, north of Cross Creek, one monarch with the A. viridis fingerprint was collected in April with a similar wing size and condition and lipid and lean mass as A. viridis monarchs from Cross Creek and Miami (Supporting Information, Table S2).

Florida is not part of the flyway for re-migrants from Mexico. Urquhart & Urquhart (1979a) tagged 33 000 monarchs at different colonies in the trans-volcanic belt of Mexico, and only one was later recaptured in Florida at Daytona Beach. This is the only known record of a monarch that was at the Mexican overwintering site and that was later recaptured in Florida (Urquhart & Urquhart, 1979a). In addition, Monarch Watch has no recoveries of monarchs from the overwintering colonies in Mexico that were later recaptured in Florida (pers. comm. A. Babbit). The fingerprints of the monarchs collected along the Florida Peninsula are a combination of six different Asclepias species’ fingerprints (Tables 2–3), in contrast to the monarchs collected at the Mexican overwintering sites where the vast majority (92%) have the A. syriaca fingerprint (Malcolm et al., 1993). These fingerprints, exhibited by the monarchs collected in Florida, are mostly the same as those shown by monarchs collected in Cuba in November which suggests that they are part of the same migratory population.

Collectively, these data show a complex migration scenario for Florida where some migrant monarchs fly south during the autumn, and some overwinter, reproduce and start moving north during the spring, a similar scenario to the Mexican overwintering monarchs. Beyond Florida, some continue their migration to Cuba, Yucatán, Guatemala and other locations through the insular Caribbean. This diversity in migration strategies exhibited by eastern North American monarchs allows them to increase the use of temporal and spatial resources that Mexican migrants do not use. In addition, the results show that, in the Americas, extensive gene flow occurs between the different monarch populations, resulting in complex genetic patterns and points to more migratory routes for the eastern North American monarchs than the well-documented one that overwinters in the oyamel forest in Michoacan.

Monarchs that are part of the south-eastern and Atlantic migratory routes are not ‘aberrant’ (Urquhart, 1987) or part of a dispersion route (Brower, 1995), but part of migratory pathways with particular routes and migration times as is supported by tagging, TLC and δ²H results. In addition, they have their own distinctive phenotype, natal grounds and genetic signature. Finally, DNA and anecdotal evidence show that some migrant monarchs return to North America and support the idea that monarchs are part of these newly proposed migratory routes and are not part of an evolutionary sinkhole.

Evolution of monarch migration

The monarch butterfly originated in southern USA or northern Mexico from a migratory ancestor around 1 100 000 years ago (Zhan et al., 2014) in an Asclepias centre known as the ‘Mexican cluster’, and from there began migrating to neighbouring centres: Ozarkian, Appalachian, Floridian, Californian and the Antillean (described in Woodson, 1954). Given the monarch’s great flying capability and that these Asclepias spp. clusters probably overlapped as they do now, we expect that the monarchs migrated between these centres and that these movements were the early origins of the different migratory routes: Mexican, south-eastern and Atlantic. However, the extensive glacial and interglacial periods, from around half a million to 12 000 years ago, drastically changed the available monarch breeding grounds and shaped the evolution of the monarch migration. During the glacial periods, the monarch breeding grounds were limited to the north by the ice sheet (roughly at 40°N during the peak of the last glaciation), and included a larger and a more connected insular and continental Caribbean in the south (Fig. 15). As a result, the eastern monarch breeding grounds were shifted in a south-easterly direction and this was probably instrumental in the evolution of south-eastern and Atlantic migratory routes that allowed an easy passage to a large portion of the Caribbean and South American coast (Fig. 15). The glacial periods were followed by interglacial periods when the monarch’s northern territory expanded and, in contrast, its southern territory contracted and became more fragmented.

Figure 15.

Portion of the Americas showing the approximate position of the shore line as a black contour and areas of shallow water in grey during the Last Glacial Maximum.

The monarch northern territory has been ice-free since the last glacial period around 12 000 years ago. This allowed expansion of both the Asclepias species and the monarch’s northern breeding grounds and explains the hundred-fold increase in the monarch’s effective population size (Zhan et al., 2014) in its northern range: Mexican and Ozarkian. In contrast, to the south, the shoreline shrunk, as the sea level rose and isolated and fragmented a large portion of the insular and continental Caribbean, which forms the southern part of the monarch’s territory. This fragmentation and isolation of the monarch territory and its population probably was the precursor of the different monarch subspecies in these areas. Another outcome was the ten-fold decrease in the effective population size of monarchs that migrated through these areas; however, for monarchs moving through south Florida, the effective population size stayed the same (Zhan et al., 2014).

The evolution of different migratory routes and strategies allowed the monarch butterfly to extensively use the Asclepias spp. present in North America.

ACKNOWLEDGEMENTS

This work took years in the making and involved family, friends and people who believe in the ideas presented, thank you all for being part of this journey to uncover the story of the monarch butterfly. We certainly hope that this helps to protect this beautiful butterfly and the natural world. We thank Amy Knight who allowed us to use her monarch data from Florida, made several of the maps, and for her input along the way. Barbara McKinnon allowed us to search her library in Merida for Yucatán monarch records and donated invaluable notes and papers of Urquhart. Roberto de La Maza read the manuscript and provided deep insight into the monarchs in Mexico and beyond. Ray Moranz helped to identify A. perennis fingerprints and Tonya Van Hook read an early version of the manuscript and offered her comments and observations. We thank David Cook and the volunteers from the St. Marks Refuge for collecting butterflies in December for this study. We would like to thank Polly Palmer for reviewing and correcting earlier versions. The editors and reviewers did a great job reviewing and editing our paper, many thanks to them. Verónica Juárez-Jaimes from the Botany Department and Laura Cortés Zárraga from the Botanical Garden at Universidad Nacional Autónoma in México (UNAM) helped us determine the Asclepias spp. present in Mexico. We have no conflicts of interest to declare. The research by supported by Environment and Climate Change Canada operating funds to K. Hobson and NSF grant IOS-1922624 to M. Kronforst.

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding author upon reasonable request.

SUPPORTING INFORMATION

Additional supporting information may be found in the online version of this article on the publisher's website.

Figure S1. Monarch flying in the northern portion of Isla Mujeres in the Yucatán on 10 May 2021 (María Cecilia Álvarez’s picture and observation). The northern portion of Isla Mujeres is only 161 km from the Cabo San Antonio, the most western portion of Cuba.

Figure S2. Observations of individual monarchs in the Yucatán in green (https://www.naturalista.mx/observations?taxon_id=48662 ) and aggregations of monarchs in San Juan del Rio of 300 individuals in March 2021, blue rectangle, and in Cacao of 100 butterflies, in early December of 2020, pink rectangle. The observations in San Juan del Rio and Cacao were made by Cecilia Alvarez, Estefania Medina, Roger Sosa, William Canto and Juan Flores. The map was made by Roger Sosa Pinto.

Figure S3. Male on the beach of Isla del Carmen, Campeche, in the Yucatán Peninsula that is apparently trying to copulate with the monarch under it. The picture was taken on 23 February 2021 by Alayola Garrido.

Figure S4. TLC of six monarchs captured in Florida, Cuba and Guatemala with a digitoxin (DIG) standard. From left, A. viridis (A.vi), A. curassavica (A.cu), A. humistrata (A.hu), A. asperula capricornu (A.as), A. perennis (A.pe) and A. syriaca (A.sy). This is a doctored image of a plate composed of different channels assembled from TLC with each channel representing an individual butterfly.

Figure S5. Generalized distribution of the Asclepias spp. identified by TLC in North America (only northern Mexico is included). For state-specific distribution see http://plants.usda.gov. Asclepias curassavica is widely distributed in Mexico (Juarez-Jimenez et al., 2007) and Guatemala (Biodiversidad de Guatema."Asclepias curassavica" (Online)https://biodiversidad.gt/portal/taxa/index.php?tid=1892&taxauthid=1&clid=0. Accessed 2 February 2013. and Margaret Dix, pers. obsv.).

Figure S6. Female collected in San Antonio Cuba on 7 March 1995. This female had the Asclepias viridis fingerprint and wing condition of 3. The probable natal ground of this female was south-eastern USA (Fig. 6).

Figure S7. Female captured in Quintana Roo on 3 March 1991. The probable natal ground of this female was south-eastern USA (Fig. 8, E12). Figure 9 shows where it was collected.

Figure S8. Female captured in Quintana Roo on 17 June 1993.The probable natal ground of this monarch was south-eastern USA (Fig. 8, E18). Figure 9 shows where it was collected.

Figure S9. Male collected in San Jose Pinula in Guatemala on 22 July 1987. This male, F10, does not appear in Figure 10 since its δ²H value was -88‰ (VSMOW) but its probable natal ground is presented in the inset map.

Figure S10. Male that looks like the Puerto Rican subspecies D. p. portoricensis, this butterfly was collected in San Antonio, Cuba, on 26 November 1997 and its probable natal ground was the south-east USA (Fig. 7, D21). This subspecies is characterized by the absence of an inner line of spots on the forewing.

Figure S11. Male that looks like the Puerto Rican subspecies D. p. portoricensis, this butterfly was collected in San Nicolas, Mexico State, Mexico on 21 November 1952. This male is in the private collection of the De la Maza y Elvira brothers in Mexico City. This subspecies is characterized by the absence of an inner line of spots on the forewing.

Figure S12. Male that looks like the Puerto Rican subspecies D. p. portoricensis, this butterfly was collected in Guatemala on 20 April 1999. This subspecies is characterized by the absence of an inner line of spots on the forewing. Picture courtesy of Jiichiro Yoshimoto, curator of entomology at the systematics laboratory of Universidad del Valle de Guatemala.

Figure S13. Male with the A. viridis fingerprint collected in Catemaco, Veracruz, Mexico in September 1964.

Figure S14. Male with the A. viridis fingerprint collected in Colima, Mexico, on 15 September 1975.

Figure S15. Phylogenetic relations among monarch populations inferred using the maximum likelihood method. This is not a fully solved tree and the colours used correspond to those in Figures 11 and 12.

Table S1. Observations of the monarchs presented in Figure 1.

Table S2. Mean ± SD of wing length (cm), wing condition, fat content (mg) and lean mass (mg) of monarchs collected in Louisiana, Georgia, five Florida locations and at overwintering colonies in Mexico (Alonso, 1996; Van Hook, 1996; Brower.], Alfonso A, 1996; Van Hook T, 1996; Brower L, 1991). For details on how wing size and condition, and fat content and lean mass, were quantified/qualified see Dockx (2012).

Table S3. Mean ± SD wing length (cm), wing condition, fat content (mg) and lean mass (mg) of female and male migrant monarchs collected at two locations in Cuba (Fig. 2). The following abbreviations were used for the localities: Guana (Guanahacabibes) and San An. (San Antonio). For details on how wing size and condition, and fat content and lean mass, were quantified/qualified see Dockx (2012).

Table S4. Eggs, five larvae instars and pupae count in Sierra Madre in Guatemala in a location north-west of Guatemala City (Fig. 9).