Tracking butterfly responses to climate change using citizen science data

Leslie RiesUniversity of Maryland, Biology

National Socio-environmental Synthesis Center

Focus on the monarch as a model for understanding climate and ecological models

Stage 1: Overwintering

Stage 2: Spring

migration and breeding

Stage 3: Summer expansion and breeding

Stage 4: Fall migration

Journey Northstarted ‘99

Monarch Larvae Monitoring

Projectstarted ‘99

WWF-Mexico• started ’96

• started tracking overwinter mortality in 2003

N. American Bfly Assoc.started 1975

Ohio BMSstarted ’96

How does climate impact butterflies? Changing climates can challenge the physiological

tolerances of species Increased heat can allow more growth (in juveniles) or activity

(in adults), but can also be harmful at excessive temperatures Changing climates can shift the distribution or emergence

timing of interacting species host or nectar plants natural enemies or mutualists

These dynamics can combine to shift range distributions or impact population numbers Species Distribution Models help us understand range dynamics Population models help us understand how performance at each

stage impacts population trajectories

• Mechanistic models translate environmental conditions (often thermal constraints to growth or energetics) into biologically relevant metrics (survivorship or fecundity) and can be used to predict distributions at large scales.

• BENEFITS: • Specific mechanisms are identified a

priori • Allows independent distribution data

to test predictions and identify specific weaknesses and strengths of the models

• DRAWBACKS:• Lack of data for most organisms • Short history of model development• Lack of model transferability between

organisms

Species Distribution Models: Mechanistic Approach

0

500

1000

1500

2000

2500

3000

3500

0 100 200 300 400Acc

umul

ated

deg

ree

days

Julian Date

Degree days through the year

Growing Degree Days (GDD)• Growing degree days are used to estimate the

amount of thermal energy available for growth. • A minimum temperature at which growth can

begin is determined (DZmin), and each degree above that is considered a “degree day”• In some cases, a maximum temperature is set

(DZmax) after which degree days are no longer accumulated

June

Apr

Oct

0

5

10

15

20

25

5 10 15 20 25 30 35 40 45

Dai

ly d

egre

e da

ys

(Tmin+Tmax)/2

Calculating daily degree days

DZmin = 11.5C

(52.7F) DZmax = 33C (91.4F)

?

Total GDD required:~323DD

Zalucki 1982

44DD

31DD

27DD

25DD

34DD

58DD

104DD

Goal: Predict current (and future) distributions

Take laboratory-measured temperature tolerances to develop GDD calculations

Obtain climate data to estimate number of generations possible within a region of interest

Determine if the predicted range overlaps with the known distribution

+ =

Building the distribution modelTake laboratory-measured temperature tolerances and intersect with spatial patterns of heat accumulation throughout eastern North America to predict number of generations that could be produced in spring and summer

Predicted number of generations on averageSpring (Mar-Apr) Summer (May-Aug)

Testing the distribution model:spring distributions

Data source: Journey North All records: Mar-Apr

Testing the distribution model:summer distributions

Data source: NABA Average observations: July

1 year

35 year

Modeling the distribution of the sachem butterfly using mechanistic models

The sachem butterfly (Atalopedes campestris) recently expanded its range into Washington state.

Winter temperatures had been rising in the area

Natural history notes: a common, open-area species that uses several grasses, including common grasses such as Bermuda and crab grass

Leslie Ries1, Jessica Turner1, Lisa Crozier2, and Thomas Mueller1

1University of Maryland 2NOAA NW Fisheries Center

Summer recruitment Recruitment predictions are based on growing degree days

In this case, the lower development threshold (DZmin) was 15.5C (60F) with no upper temperature threshold (Crozier 2001)

Number of growing degree days necessary for a single generation was estimated at 417

Predictedgenerations:

Predicted number of generations based on NOAA weather station data (1990-2009 average)

Overwinter survival Predicted rates of overwinter survival are based on mean

January temperature

Predicted overwinter survival based on NOAA weather station data (1990-2009 average)

Predictedsurvival:

Predicted growth rates The model predicts lambda based on summer recruitment

and overwinter survival Clearly, the scaling of the model predictions seem off…

Predicted lambda based on NOAA weather station data (1990-2009 average)

Predictedlambda:

The model does well at capturing the northern range limits (limited by cold), but

overestimates growth in warmer regions

GDD models have historically focused on cold limitations rather than physiological detriments of excessive heat

Predictedlambda:

Observed abundances:

Another study showing the sachem increasing in MA

*

GDD models have historically focused on cold limitations rather than physiological detriments of excessive heat

Lethal and sub-lethal temperature effects were tested in a laboratory

setting (Betalden et al., in prep) Larvae at various stages were exposed to potentially

lethal or sublethal temperatures for a different number of days 38C (100.4F), 40C (104F), 42C (107.6F), 44C (111.2F) and

a control (30C, 86F) First, Third, Fifth instars exposed Exposed for 1, 2, 4, or 6 days Nighttime temperatures were kept at 25C

Larvae were reared to determine survivorship rates and total development time (in degree days)

Results: Survivorship rates (Betalden et al., in prep)

• Survivorship begins to decline for the 40C treatment only when larvae are exposed for 6 days (and only for 3rd and 5th instars)

• Survivorship is very low overall at 42C

• No individuals in the 44C treatment survived

Results: Development Time (Betalden et al., in prep)

• Development time increases as individuals are exposed to higher temperatures for longer periods of time

• There is a treatment effect even for individuals exposed for 1 day (suggesting sublethal effects may occur at lower temperatures)

So how should GDD be calculated?

0

5

10

15

20

25

5 10 15 20 25 30 35 40 45

Daily

deg

ree

days

(Tmin+Tmax)/2

Calculating daily degree days

DZmin = 11.5C (52.7F)

DZmax = 31.5C (88.7F)

?

Sub-lethal and lethal

effects

?

Mapping out lethal and sub-lethal zones:Average from 1990-2009

ABOVE 42ABOVE 40ABOVE 38

AVG NUMBER OF DAYS

• Lethal and sub-lethal temperatures seem to correspond to limits in population densities, especially in the midwest

• Next up:• Comparing lethal temperatures to

larval development observed in the field

MLMP SITES

Preference and performance relative to mean number of days >38C

Mean number of days with temps >38C

0

0.2

0.4

0.6

0.8

1

June July Aug

Prop

orti

on E

ggs

Month

Performance

0-1

1-5

>5

In zones where caterpillars are more likely to experience sub-lethal temperatures, there appears to be less development

But climate differs from year to year – so did temperature really drive those patterns?

Accumulated sub-lethal degree days:

1999 2000 2001

2002 2003 2004

• These are accumulated over the main summer growing season (2 months)

• To truly test the impacts of sub-lethal and lethal temperatures, we need to tie temperature events to survey dates

Relationship between development and accumulated sub-lethal degree days

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

None 0-1 1-10 >10

Prop

orti

on la

te in

star

larv

ae

Number of accumulated sub-lethal degree days

n=518

n=267n=70

n=24

• Conclusions• Monarch distributions are limited by cold temperatures in the

spring and both cold and hot temperatures in the summer• But what stage is the most crucial for monarch populations?

Understanding monarch population dynamics is critical for their conservation

Notable population patterns: Eastern monarchs may be declining,

but examining different life stages suggests different patterns

Monarch populations show large fluctuations from year to year

Underlying mechanisms REGIONAL CONNECTIONS: How

do dynamics in one phase of the migratory cycle influence dynamics in subsequent phases?

ENVIRONMENTAL INFLUENCE: How much do environmental factors influence the connection between these phases?

SUMMER MONITORING DATA

WINTER MONITORING DATA

Focus on climate impacts during migration and breeding

Stage 1: Overwintering

Stage 2: Spring

migration and breeding

Stage 3: Summer expansion and breeding

Stage 4: Fall migration

Journey Northstarted ‘99

Monarch Larvae Monitoring

Projectstarted ‘99

WWF-Mexico• started ’96

• started tracking overwinter mortality in 2003

N. American Bfly Assoc.started 1975

Ohio BMSstarted ’96

Data Available for Analysis

Monarch Larvae Monitoring Project (MLMP)

North American Butterfly Association Counts (NABA)

Ohio Butterfly Monitoring Scheme (OH)

Mexican sites

WWF: Mexico

Data Available for Analysis

Monarch Larvae Monitoring Project (MLMP)

North American Butterfly Association Counts (NABA)

Mexican sites

South

N-Central

N-East

Ohio Butterfly Monitoring Scheme (OH)

WWF: Mexico

South

N-Central

N-East

Tracking the population through each region and stage

1. Do the number of adults surviving the winter in Mexico relate to the number of adults arriving in the

Texas area in spring?2. Do the number of spring arriving

adults relate to the number of 1st gen eggs that are recorded?

Mexican sites

South

N-Central

N-East

Tracking the population through each region and stage

Mexican sites

1. Do the number of adults surviving the winter in Mexico relate to the number of adults arriving in the

Texas area in spring?2. Do the number of spring arriving

adults relate to the number of 1st gen eggs that are recorded?

3. How do the number of spring adults or eggs relate to the number of 1st generation arrivals in the northern

regions?

South

N-Central

N-East

Tracking the population through each region and stage

Mexican sites

1. Do the number of adults surviving the winter in Mexico relate to the number of adults arriving in the

Texas area in spring?2. Do the number of spring arriving

adults relate to the number of 1st gen eggs that are recorded?

3. How do the number of spring adults or eggs relate to the number of 1st generation arrivals in the northern

regions?4. Can the number of 1st generation

adults / 2nd generation eggs predict numbers in subsequent generations?

Q1 and Q2. How do overwintering numbers relate to the number of arriving adults and how do arriving adults influence the

number of eggs we see in the spring?

0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3

2nd

gen

MLM

P eg

g de

nsit

y (N

-Cen

tral

)

1st gen MLMP egg density (South)

00.20.40.60.81

1.21.41.61.82

0 0.5 1 1.5 2 2.5N

ABA

1st

Gen

adu

lts (

N-E

ast)

NABA spring migrants (South)

Q3. How do the number of spring adults or eggs relate to the number of 1st generation arrivals in the northern regions?

A weak, or non-existent, relationship between the spring generation and summer arrivals in the north could be due to lack of data, or swamping out by environmental

factors.

1st gen eggs in south to 2nd gen eggs North Migrant adults in south to 1st gen adult arrivals

r=0.5 p=0.13 r=0.69 p=0.12

0

1

2

3

4

5

6

7

8

0 1 2 3

Sum

mer

NA

BA d

etec

tions

(N-C

entr

al)

1st Gen Spring arrivals NABA (N-Central)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.01 0.1 1

Sum

mer

MLM

P eg

g de

ns (N

-Cen

tral

)

2nd gen MLMP egg density (N-Central)

Q3. Can the number of 1st generation adults / 2nd generation eggs predict numbers in subsequent generations?

YES: This suggests that the number of arrivals in the northern breeding grounds from the southern spring generation has a strong influence on the ultimate size of that year’s population.

0

1

2

3

4

5

6

0.001 0.01 0.1 1Su

mm

er N

ABA

det

ectio

ns (N

-Eas

t)2nd gen MLMP egg density (N-East)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.001 0.01 0.1 1

Sum

mer

MLM

P eg

g de

nsit

y (N

-Eas

t)

2nd gen MLMP egg density (N-East)

N-East: r=0.899 p=<0.0001 r=0.85 p=0.004

r=0.7925 p=<0.0001 r=0.73 p=0.005

N-Central:

Tracking climate’s impacts on the migratory monarch butterfly

We examined the impacts on population growth in Ohio of:

1. Spring temperature (in Texas)2. Spring precipitation (in Texas)3. Summer temperature (in Ohio)

4. Summer precipitation (in Ohio)

Patterns based on simple climate metrics aren’t informative

Meaningful patterns emerge when patterns are evaluated in a multiple regression framework

The story so far… • No relationship between adults leaving Mexico, arriving in the South, and

laying eggs

• Weak (or non-existent) relationship between adults arriving in the South, next generation arrivals in the North

and egg-laying• The disconnect may be due to the importance of spring climate on the

ultimate population size (and/or health) of migrants to the North

• A strong relationship between the numbers arriving in the North and

laying eggs and the size of the population at the end of the summer.• This suggests that the size of that

first generation produced in the spring that arrives in the North is an

important contributor to yearly population sizes and (again) that

spring climate is important

But are climate impacts direct or indirect?

South

N-Central

N-East

Could climate impact the phenological linkage between milkweeds and monarchs

Data source: Journey North

Arrival dates - 2001

Arrival mismatches are leading to at least anecdotal incidents of egg loading

Take home messages

Climate is impacting butterfly populations Using models that link physiological tolerance data to

large-scale distributions is a powerful way to tease out the complex interactions between climate and ecology

We could not possibly explore these questions in a rigorous way without a data stream from dedicated citizen-science networks, so public participation in scientific research is CRUCIAL to answer the most relevant questions in today’s world.

So THANK YOU!!!

Questions?