Monarch Watch Blog

Monarch population crash in 2013

Friday, June 11th, 2021 at 10:27 am by Chip Taylor
Filed under Monarch Biology | Comments Off on Monarch population crash in 2013

What contributed to the monarch population crash in 2013?

The text below was written at the time the U.S. Fish and Wildlife Service was evaluating evidence of all aspects of the monarch population prior to determining whether the monarch should be listed as a threatened species. As you may remember, in December 2020, monarchs were listed as threatened but precluded based on the priorities given to other species which were more clearly threatened and endangered. A final decision on the status of monarchs is scheduled for 2024. The below was neither published nor posted to this blog. It should have been since the analysis is relevant to what happened between 2016-2018 in California that led to the collapse of the Western monarch population. I’ve reorganized and added to the original text to provide more background and clarity.


There are numerous ways to associate trends in the growth or declines in populations with physical (weather related) and biological factors. Ideally, there are real numbers to work with but that is seldom the case, and lacking hard counts, we often have to link outcomes tentatively with causative factors until we see the patterns appear over and over again in generation to generation or year to year records. That’s the case with monarchs. Numbers are generally lacking with cause-and-effect relationships first appearing as speculation or inference. Those insights, if confirmed again and again, can morph into hypotheses, and in the best cases, correlations, which, in themselves, do not guarantee causation. The next step is prediction. If the outcomes can be predicted based on the observed relationships, we can be reasonably sure we are on the right track.

My approach to trying to understand the trends in monarch numbers has been to use what is called a stage specific model. I break the annual cycle into 6 stages based on the dominant activity within a stage. Each stage is treated as a discrete period that demographically represents either an increase or decrease in numbers depending on conditions during the period but not excluding the condition of the butterflies entering a stage. For example, drought starved monarchs could arrive at the overwintering sites with shorter wings, smaller mass and reduced fat bodies that could affect both mortality during the stage and realized fecundity of the survivors. Further, each stage sets the initial numbers for the next stage, and, in effect, provides limits on what can happen in the next stage irrespective of the conditions during that stage. In other words, the outcome of the next stage depends both on the starting numbers and fitness (mass, fat body, wing area) of the starting population and the conditions during that stage. Good starting numbers may not result in an expected increase if conditions are poor, and conversely, poor starting numbers can effectively cap the outcome even under the most favorable conditions. It also appears to be extremely difficult for a population to recover from a poor start in a breeding season that only involves 3-4 generations.

This approach evolved from attempts to understand the variation in year-to-year measurements of the size of the overwintering monarch population in Mexico. My quest started with a simple question: what was the explanation for the decline from 18.19 hectares in the winter of 1996-1997 to the following winter of 1997-1998 with a population of 5.77 hectares? The answer appears to be an April frost in Texas that killed large numbers monarch eggs and larvae. That quest led to an analysis of the factors possibly associated with population growth for all years from 1994 forward as well as searches through the weather records back to 1895. The point I’m making is that, although this text focusses of what happened from 2011-2015, the data base I’ve used to interpret these records is extensive.


The metrics used in this analysis are large scale, or reflect measures of large-scale, that is multi-regional, events. Regional weather and local and regional impacts due to predators, parasites and pesticides all have a role in determining population numbers but are difficult to measure and hard to integrate into a stage specific model. With this in mind, I have selected only a few measures that appear to be correlated with outcomes in a way that give them predictive value. Among these measures are first sightings of adult monarchs in the spring, March temperatures in Texas, May temperatures north of 40N, summer temperatures and NDVI, a measure of greenness of the vegetation on a landscape (Table1). Success or failure of population growth is measured by the increase or decrease in the areas occupied by monarchs at the overwintering sites from one year to the next.


The population declined from 2011 to 2012 and again from 2012 to 2013 and increased in both 2014 and 2015 (Table 1). The declines occurred as the result of high March temperatures in TX (2011, 2012), summer temperatures >1.9F above the long-term means (2011, 2012), drought conditions (low NDVI) in the South Region (2011, 2012) and a low number of first sightings (2013). The increases in 2014 (0.67 to 1.13) and 2015 (1.13 to 4.02) followed favorable mean temperatures for population growth in March, May and June-August. The NDVI index, which can be taken as a proxy for the availability of nectar sources, was high for each year as well

The associations of high mean temperatures and drought conditions with population decreases is seen throughout the record from 1994 through 2019. Similarly, population growth from one year to the next is associated with mean temperatures that are close to or below the long-term average for March in TX. The longer record shows that populations increase when summer means are up to 1.9F above long-term means but decrease when mean temperatures substantially exceed +1.9F. Populations also decline when mean temperatures are >-1.5F below average (2004, 2009).

The temperatures for September and October represent the conditions for the first and second halves of the fall migration (Table 1). Although there is no clear relationship between these means and the size of the overwintering population in these data, this may be changing. The extreme high temperatures for September 2019 delayed the migration by up to two weeks. Those temperatures and a drought in Texas and northern Mexico resulted in an unusually low recovery rate (N=392) for tagged monarchs suggesting that attrition due to these factors limited migratory success. Similarly, above average September temperatures in the Northeast in recent years are associated with lower recovery rates of tagged butterflies from this region.


The low number of monarchs reported at the overwintering sites in Mexico in the winter of 2013-2014 appears to have been the result of a series of negative weather events that began in the summer of 2011. Excessive temperatures and droughts in 2011 and 2012 followed by low initial colonizing numbers in the spring of 2013 account for the decline. More favorable conditions for population growth allowed the population to increase in 2014 and 2015.


Summer temperatures can have a significant impact on the size of the fall migratory population. A review of all the records suggests that population growth is enabled when temperatures range from a little over -1F to almost +1.9F. Lower temperatures of >-1.5F, as in the summers of 2004 and 2009, both due to a southward dip in the jet stream, resulted population declines. These reductions may have been due a reduction in egg development which is temperature dependent, egg laying rate or simply a shorter growing season or all three. Average temperatures in the Upper Midwest greater than +1.9F were also associated with declines. Such temperatures are likely to negatively affect realized fecundity by reducing the reproductive activity and longevity of adults. In addition, such temperatures affect resources available to adult monarchs by shortening flowering intervals, and hastening senescence of host plants.


A metric known as normalized difference vegetation index (NDVI), a measure of greenness derived from satellite imagery, is frequently used to measure the severity of droughts. Low values indicate severe drought conditions and the NDVI values for Texas were low in both 2011 and 2012. These low values are associated with low recoveries of tagged monarchs and lower than expected overwintering numbers. Drought conditions also occurred in some parts of the Upper Midwest in 2012. Those conditions may have further reduced the migration and the size and robustness of the butterflies originating from that region. Reports of “small” monarchs occur with some frequency during droughts in the Midwest.

First sightings

While first sightings posted to Journey North tend to be urban and suburban centric and can be limited by the number of people willing to report sightings, they appear to tell us two things that are relevant here. First, low numbers, well below the recent average numbers of sightings, as in 2013, tell us that population growth can be limited by the number of returning monarchs. This record also tells us that there is no obvious association between the size of the overwintering population and the number of first sightings. For example, while the overwintering population of 2012 (1.19 hectares) led to 322 first sightings in 2013, the lower population of 2013 (0.67 hectares) accounted for 521 first sightings in 2014. Similarly, the larger population of 2014 (1.13 hectares) produced only 547 first sightings. Sorting out the human factor from the dynamics that produce the numbers of monarchs moving north in March and later in May will be a challenge. In the meantime, the low number returning in 2013 gives rise to several questions about the fitness of the cohort of monarchs that arrived at the overwintering sites in the fall of 2012. We know that most monarchs originate from the Upper Midwest and that the region experienced a semi-drought that summer along with the highest temperatures in the record for that region from 1994 to present. The tagging record for 2012 indicated that numbers tagged that fall were among the lowest we’ve recorded for the Upper Midwest. Therefore, it’s possible that the last generation monarchs were less fit due to lower fat reserves, or size, to survive the migration, the winter period and the migration northward the following spring. High levels of mortality during the migration northward from the colony sites to Texas could also account for the low number of first sightings in the spring of 2013.

Additional considerations

There are a few more things to say about 2012 and 2013 that are not apparent from the above data. The first sightings data show that the return migration for 2012 was the earliest in the record with large numbers of overwintered monarchs advancing beyond 40N in late April and the first 10 days of May. In contrast, due to late arrivals and cool conditions in May of 2013, the movement of first-generation monarchs beyond 40N was predominantly at the end of May. These two back-to-back years demonstrated that overwintering and first-generation monarchs can advance both too early and too late into the summer breeding area for optimal population development. If too early, females lay eggs at more northerly latitudes with cooler, sometimes freezing, temperatures that delay age to first reproduction. Such delays have the effect of reducing the reproductive success of the returning cohort that overwintered in Mexico. Late recolonization also negatively affects population growth in that it shortens the breeding season.

This analysis suggests that it is extremely difficult, maybe impossible, for a population to recover from a poor start in a 6.5-month breeding season that involves 3, and more rarely 4, generations. Of course, this depends on the definition of a poor start, and in this case, I’m thinking of the poor starts in 2004, 2012, 2013. There may have been carry-over effects from the droughts of 2002 and 2012 that had an impact on the populations the following years. The population that developed in 2012, following the drought of 2011 also declined, but that decline may have been due as much to conditions in March and August as to the condition of the butterflies surviving from the previous winter.

This example shows that the decline and recovery of a population can be explained IF measures can be identified that have a broad geographic impact on a population at a particular stage. Negative events can have a ripple effect from one stage to another and can carry over from one year to the next. Further, a series of negative events can cause a population to spiral downward rapidly. Is it possible that the downward trends in the Western monarch population from 2016-2020 were the result of a similar series of negative events? Yes, that’s possible.

The crash in the population numbers in 2013, together with what appeared to be a long-term decline that included other years with alarmingly low numbers (2009-2010 – 1.92 and 2012-2013 – 1.19) led to the petition filed in August 2014 to declare the monarch a threatened species. This petition was preceded by an invitation from the White House to a stake-holders meeting at the Eisenhower Office Building in April 2014. Attendees represented a broad array of organizations interested in monarch and pollinator conservation. A bit later, on the 20th of June 2014, President Obama issued a memorandum requesting that all Federal Agencies with some control of landscapes become engaged with monarch and pollinator conservation. Given the prospect that a small overwintering population would be extremely vulnerable to catastrophic overwintering mortality that could reduce the numbers to levels that would threaten the very existence of the monarch migration, these responses were not unreasonable at the time. The U.S. Geological Survey, the lead research agency of US government, took the lead in convening meetings to examine the monarch situation. These meetings were populated with monarch biologists and appropriate expert personnel from federal agencies. A number of publications resulted from these meetings, the most cited of which is “The all hands-on-deck” paper (Thogmartin, et al, 2017). The analysis therein indicates that an overwintering population of 6 hectares is required for monarchs to be able to rebound from known adverse weather or other events. The analysis also points out that to achieve a population of this size will require the restoration of 1.8 billion milkweed stems. While there are many restoration efforts under way, it’s not clear whether these efforts are sufficient to offset the 2 million or more acres of potential monarch habitat that are converted to croplands and lost to development each year.

The low overwintering numbers of 2013-2014 led me to ask whether monarch numbers had been this low or lower in the past. Yes, that is likely, but there is no data on this subject. What we can do is look at past climates. Those records show that monarchs have experienced far greater extremes in the past than anything seen since 1994. We can only speculate about the impact of these events on monarch numbers, but it seems likely monarchs declined to extremely low levels several times in the last 125 years. And before that was the “little-ice-age”, a much colder period with erratic shifts in climate from 1300 to 1850. While monarchs clearly survived numerous extremes through the centuries, they now face new threats posed by climate change such as severe winter storms, high temperatures and droughts in all portions of the breeding season and even the migration. Those impacts can already be seen in California.

Table 1. Temperature means, first sightings and NDVI records for 2011-2015.

Overwinter hectares/prev season4.022.891.190.671.13
First sightings
1 March–30 April
March mean temp TX+5.4+6.8+0.9-1.6-0.2
May mean temp 40N UM-0.2+4.8+0.2+0.5+1.3
First sightings >40 N
1 May–9 Jun
June–August mean temp
Upper Midwest
September mean temp
Upper Midwest
October mean temp
South Region
Total first sightings603663322521547


Wayne E Thogmartin, Laura López-Hoffman, Jason Rohweder, Jay Diffendorfer, Ryan Drum, Darius Semmens, Scott Black, Iris Caldwell, Donita Cotter, Pauline Drobney, Laura L Jackson, Michael Gale, Doug Helmers, Steve Hilburger, Elizabeth Howard, Karen Oberhauser, John Pleasants, Brice Semmens, Orley Taylor, Patrick Ward, Jake F Weltzin and Ruscena Wiederholt. (2017). Restoring monarch butterfly habitat in the Midwestern US: ‘all hands on deck’. Environmental Research Letters, Volume 12, Number 7 074005

Sorry, comments for this entry are closed at this time.