This document summarizes a study on developing a habitat distribution model for the seaweed Himanthalia elongata in response to warming climate conditions. The study collected presence/absence data for H. elongata across 1,189 grid cells and selected 7 environmental predictors including sea surface temperature, air temperature, cloudiness, waves, and substrate type. Statistical analyses identified the most important predictors as mean August sea surface temperature, mean maximum August air temperature, and presence of rocky substrate. The results indicate H. elongata is currently present in colder sea and air conditions, suggesting its distribution may retract westward as temperatures rise with climate change.
Habitat Model of Himanthalia elongata: responses to a warming climate
1. Habitat Model of Himanthalia elongata:
responses to a warming climate
Brezo Martínez, Rosa M. Viejo,
Francisco Carreño & Silvia C. Aranda
brezo.martinez@urjc.es
Universidad Rey Juan Carlos
Madrid, Spain
2. SPECIES
DISTRIBUTION MAJOR RESEARCH
TOPIC DURING
LAST DECADES
BIOGEOGRAPHY
Grinnell 1917 Am. Nat.
ENVIRONMENTAL Gaston 2003
Gaston 2009 Proc. R. Soc. B.
DRIVERS
3. SPECIES
DISTRIBUTION HUMAN IMPACTS:
- Habitat fragmentation
BIOGEOGRAPHY - Alien species
- Biodiversity crisis
- Global Climate Change
ENVIRONMENTAL
DRIVERS
Parmesan et al. 2005 Oikos
4. IPCC, 2007
GLOBAL WARMING RANGE SHIFTS
Species worldwide: Higher altitudes
Pole-ward displacement
Thomas & Lennon 1999 Nature
Wilson et al. 2005 Ecol. Lett.
Parmesan 2006 Annu. Rev. Ecol. Evol. Syst.
5. IPCC, 2007
Spain: sun, siesta…
PREDICTIONS: high thermal increase
6. INCREASING NW Iberia
OCEAN TEMPERATURES
0.1 - 0.5 ºC per decade since mid 70s
deCastro et al. 2009 Cont. Shelf Res.
Michel et al. 2009 Cont. Shelf Res.
7. INCREASING NW Iberia
OCEAN TEMPERATURES
0.1 - 0.5 ºC per decade since mid 70s
deCastro et al. 2009 Cont. Shelf Res.
Michel et al. 2009 Cont. Shelf Res.
FIRST EVIDENCES OF RETRACTIONS OF
COLD-TEMPERATE INTERTIDAL SEAWEEDS:
Himanthalia elongata: 130 km W Cantabrian Sea -N Spain
Fernández & Anadón 2008 Algas
15. AIMS
PREDICTIVE MODEL EQUATION
for Himanthalia elongata distribution
- Relate occurrence/absence with:
Oceanic & atmospherical climatic variables
Non-climatic variables & 2D resources
- Predict trends in the ongoing Global Warming
16. GIS + HABITAT DISTRIBUTION MODELS
Fully developed in terrestrial systems
Guisan & Zimmermann 2000 Ecol. Model.
Austin 2002 Ecol. Model.
Heikkinen et al. 2006 Prog. Phys. Geog.
To our knowledge…
first HABITAT MODEL for an intertidal seaweed
18. Many STATISTICAL TECHNICHES
available: multiple regression, logistic
regression - GLMs, GAMs, HP, AIC, BIC, etc.
BIOLOGICAL
DATA
sp1
Presence/absence sp1
Abundance sp1
Herbarium data
19. Many STATISTICAL TECHNICHES
available: multiple regression, logistic
regression - GLMs, GAMs, HP, AIC, BIC, etc.
BIOLOGICAL ENVIRONMENTAL
DATA PREDICTORS
sp1 X, Y, Z,
Presence/absence sp1 Climatic variables:
Abundance sp1 temperature, humidity
Herbarium data Spatial variables:
kind of substrate
20. Many STATISTICAL TECHNICHES
available: multiple regression, logistic
regression - GLMs, GAMs, HP, AIC, BIC, etc.
BIOLOGICAL ENVIRONMENTAL
DATA PREDICTORS
sp1 X, Y, Z,
Model equation (multivariate Logistic regression): p, the
probability of occurrence of the sp1
e a + bx + zy..
P=
1 + e a + bx + zy..
21. POWERFOOL TOOL:
- MAP of the probability of presence of the sp1 in
areas not sampled, DESCRIPTIVE TOOL
- HYPOTHESIS TESTING on species distribution-
environmental variables relationships
- PREDICTIVE EQUATION, trends in future
scenarios
22. Himanthalia elongata
BIOLOGICAL DATA:
Absences/presences
in 198 1 Km2 UTM cells.
23. Himanthalia elongata
BIOLOGICAL DATA:
Absences/presences
in 198 1 Km2 UTM cells.
1. Fucal.
24. Himanthalia elongata
BIOLOGICAL DATA:
Absences/presences
in 198 1 Km2 UTM cells.
1. Fucal.
2. Biannual.
25. Himanthalia elongata
HIGH TIDE
BIOLOGICAL DATA:
Absences/presences
in 198 1 Km2 UTM cells.
LOW TIDE
1. Fucal.
2. Biannual.
3. Low intertidal.
27. Himanthalia elongata
BIOLOGICAL DATA:
Absences/presences
in 198 1 Km2 UTM cells.
Canopy forming
Seasonally dominant
Autogenic engineering
TRACK CHANGES IN THE
WHOLE COMMUNITY
28. Himanthalia elongata
BIOLOGICAL DATA:
Absences/presences
in 198 1 Km2 UTM cells.
Cold temperate species:
southernmost limit
in N Portugal
and mid N Spain
29. Winter Spring
Distribution along the coldest coast line
(Lima 2007, Araújo et al. 2009).
Summer Autumn
Figure 2. The 21-year (1985–2005) mean sea surface temperature (SST) (C) field for (a) winter,
(b) spring, (c) summer, & (d) autumn. From Goméz-Gesteira et al. 2008.
33. 7 ENVIRONMENTAL PREDICTORS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
34. 7 ENVIRONMENTAL PREDICTORS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
35. 7 ENVIRONMENTAL PREDICTORS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
36. 7 ENVIRONMENTAL PREDICTORS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
THERMAL SEA & AIR
CONDITONS
37. 7 ENVIRONMENTAL PREDICTORS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
SURROGATE OF LOW TIDE CONDITONS
38. 7 ENVIRONMENTAL PREDICTORS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
IMPORTANT FOR ALGAE FIXATION & SURVIVAL
39. 7 ENVIRONMENTAL PREDICTORS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
SOURCES:
- SSTs & cloudiness: NOAA (US National Oceanic &
Atmospheric Administration) family satellites (2002 to 2004).
- Air Temperatures: WorldClim database (Hijmans et al. 2005).
- Waves: Coastline Modelling System Software (SMC 2.0), Coasts
& Oceanographic Engineering Group, University of Cantabria.
- Substrate: topographic maps & aerial pictures.
40. RESULTS:
1. Mean August Sea Surface Temperature SST AUG
2. Mean February Sea Surface Temperature SST FEB
3. Mean maximum air temperature of August MaxT AUG
4. Mean minimum air temperature of February MinT FEB
5. Mean August cloudiness CLOUDINESS
6. Mean wave height WAVES
7. Presence/absence of rocky substratum ROCK
VARIABLES FINALLY
INCLUDED IN THE MODEL.
41. 40
Prese
20
RESULTS:
Bifurcaria bifurcata
Himanthalia elongata B
A
100
Presences (%)
Presences (%)
80
60
40
40
20
20
Himanthalia elongata A 14.3 14.9 15.5 16.1 16.7 17.3 17.9 18.5 19.1 19.8 20.4
100 Presences Bifurcaria bifurcataSST (º C)
AUGUST MEAN B
Percent (%)
80 Absences
100
Presences (%)
60 80
40 60 Himanthalia elongata
20 80
40
Presences (%)
20
Bifurcaria bifurcata ROCK
NO B 60
14.3 14.9 15.5 16.1 16.7 17.3 17.9 18.5 19.1 19.8 20.4
100
40
Percent (%)
80 AUGUST MEAN SST (º C)
60
40 21.6 22.1 22.7 23.2 23.8 24.3 24.8 25.4 25.9 26.5
20
MAXIMAL AIR TEMPERATURE
(August mean, ºC)
Pelvetia canaliculata C
100
t (%)
80
42. 40
Prese
20
RESULTS:
Bifurcaria bifurcata
Himanthalia elongata B
A
100
Presences (%)
Presences (%)
80
60
40
40
20
20
Himanthalia elongata A 14.3 14.9 15.5 16.1 16.7 17.3 17.9 18.5 19.1 19.8 20.4
100 Presences Bifurcaria bifurcataSST (º C)
AUGUST MEAN B
Percent (%)
80 Absences
100
Presences (%)
60 80
40 60 Himanthalia elongata
20 80
40
Presences (%)
20
Bifurcaria bifurcata ROCK
NO B 60
14.3 14.9 15.5 16.1 16.7 17.3 17.9 18.5 19.1 19.8 20.4
100
40
Percent (%)
80 AUGUST MEAN SST (º C)
PRESENT IN CELLS
60
WITH ROCKY SUBSTRATUM
40 21.6 22.1 22.7 23.2 23.8 24.3 24.8 25.4 25.9 26.5
20
FREE OF SEDIMENTS MAXIMAL AIR TEMPERATURE
(August mean, ºC)
Pelvetia canaliculata C
100
t (%)
80
43. 40
Prese
20
RESULTS: PRESENT IN COLD SEA &
AIR CONDITIONS B
Bifurcaria bifurcata
Himanthalia elongata A
100
Presences (%)
Presences (%)
80
60
40
40
20
20
Himanthalia elongata A 14.3 14.9 15.5 16.1 16.7 17.3 17.9 18.5 19.1 19.8 20.4
100 Presences Bifurcaria bifurcataSST (º C)
AUGUST MEAN B
Percent (%)
80 Absences
100
Presences (%)
60 80
40 60 Himanthalia elongata
20 80
40
Presences (%)
20
Bifurcaria bifurcata ROCK
NO B 60
14.3 14.9 15.5 16.1 16.7 17.3 17.9 18.5 19.1 19.8 20.4
100
40
Percent (%)
80 AUGUST MEAN SST (º C)
PRESENT IN CELLS
60
WITH ROCKY SUBSTRATUM
40 21.6 22.1 22.7 23.2 23.8 24.3 24.8 25.4 25.9 26.5
20
FREE OF SEDIMENTS MAXIMAL AIR TEMPERATURE
(August mean, ºC)
Pelvetia canaliculata C
100
t (%)
80
47. ABSENCE AREA &
N Portugal LIMIT
High maximal atmospheric
temperature
48. RESULTS:
Himanthalia elongata
p (y) = EXP (19.40 + 1.50 * ROCK - 0.46 * MAXAGO - 0.61 * SSTAGO)
(1 + EXP (19.40 + 1.50 * ROCK - 0.46 * MAXAGO - 0.61 * SSTAGO)
Absence / presence
1
PREDICTIVE MODEL EQUATION
0
0 0.5 1
Prevalence: 0.34
Relative true presences: 69%, relative true absences: 71%(y)
Predicted probability of presence - p
AUC = 0.767, Kappa = 0.38 - FAIR discrimination
Predicted P with Jack-knifing procedures and with full data set: R2= 0.99
49. WARMING SCENARIOS (increases per decade):
- SCENARIO 1- MORE CONSERVATIVE
SST AUG 0.37 ºC Cantabrian Sea
0.27 ºC Atlantic coast
MaxT AUG 0.34 ºC until 2040 then 0.69 ºC
Sea: Gómez-Gesteira et al. 2008 J. Geophys. Res.-Oceans.
Air: B1 low emission scenario, Spanish Environment Ministry
- SCENARIO 2- LESS CONSERVATIVE
SST AUG 0.53 ºC Cantabrian Sea
0.50 ºC Atlantic coast
MaxT AUG 0.52 ºC until 2040 then 1.21 ºC
Sea: trend by Llope et al. 2006 J. Geophys. Res. Oceans.
Cabanas et al. 2003 ICES Marine Science Symposia
Air: A1 high emission scenario, Spanish Environment Ministry
50. Both scenarios underestimate the
rate of August changes:
- Reported warming trends for SST
- Annual Mean SST instead of SST AUG
- Mean Summer Air instead of MaxT AUG
Absence of georeferred future climatic trends at the studied scale.
GENERAL TRENDS!
51. RESULTS:
Probability of presence = 0 in all cells within the
CANTABRIAN SEA - extinction:
THERMAL ANOMALIES:
1.3-1.4 ºC – Mean August SST
1.5-1.4 ºC – Mean Maximal Air Temperature of August
36 and 27 years for the 1st and 2nd scenarios
… AND 58 or 38 YEARS LATER, EXTINCTION FROM THE
ENTIRE IBERIAN PENINSULA
53. CONCLUSIONS
The model suggests a southern limit associated
to OCEAN and ATMOSPHERICAL hot conditions
during summer.
First time the contribution of
both modeled
versus ISOTHERMS MODELS
54. CONCLUSIONS
The model predicts the extinction of
Himanthalia elongata from the Cantabrian Sea first
since departing ocean temperatures are harsher
and increasing at a faster rate than in the Atlantic
front (and coef. higher in the equation).
Different scheme N Spain versus N Portugal
55. CONCLUSIONS
Our results are supported by the current
retraction of Himanthalia elongata in the
Cantabrian Sea reported by Fernández &
Anadón 2008 Algas during the ongoing
warming period.
57. CONCLUSIONS
… faster than predicted.
SUPERIMPOSED
SHORT FLUCTUATIONS
Goikoetxea et al. 2009 Cont. Shelf Res.
Michel et al. 2009 Cont. Shelf Res.
In situ fixed HOBO TidbiTs: coastal sea
Temperature increase >2.8 ºC 2005-2008
58. CONCLUSIONS
We are facing a major constrain of cold
temperate flora in the Iberian Peninsula in
the ongoing warming
64. MORE WORK ON THIS:
- Demography marginal vs. central populations (Viejo et al.
2011 Ecography).
- Physiological experiments manipulating low tide x
submersed stress (1 Spanish and 1 Portuguese founded
projects). Producing a MECHANISTIC DISTRIBUTIONAL
MODEL.
- Cascade effects in the intertidal community of the
disappearance of canopy algae – diversity & functioning,
trophic structure (1 Spanish founded project)
66. Habitat Model of Himanthalia elongata: responses to a warming climate.
Brezo Martínez*, Rosa M. Viejo, Silvia C. Aranda & Francisco Carreño.
Universidad Rey Juan Carlos, Tulipán s/n, 28933 Móstoles, Spain.
* brezo.martinez@urjc.es
Traditional models explaining macroalgae distribution are based on ocean temperature
(August and February Isotherms). The predictive power of this approach is restricted
because it lacks a mathematical expression, and omits factors relevant for intertidal
seaweeds distribution, particularly atmospheric temperature. We aimed to improve our
understanding about the role of other factors in seaweed biography using a niche-
based approach suitable to make projections. Presence and absence records of
Himanthalia elongata (Linnaeus) S.F.Gray from 1 km2 cells selected on a grid along the
NW Iberian Peninsula were related using GLMs and HP to the variables that potentially
influence this species distribution: August and February SSTs, maximum and minimum
air temperature, kind of substrate, wave height, and cloudiness. Three of the former
were included in the final predictive equation: 1. presences were restricted to places
with hard substrate for fixation, 2. absences were found in sites with high August SST
at the mid Cantabrian Sea, 3. hot conditions during low tide i.e. maximum august
atmospheric temperature, restricted presences at the south, largely explaining the N
Portuguese limit. The equation was applied to two warming scenarios suggesting
Himanthalia elongata extinction from the Iberian Peninsula in few decades in response
to both increasing ocean and atmospheric temperatures. In fact, a retraction of this
species border of about 130 km has been recently noted. This confirms the model
projected trend, although it is occurring at a faster rate. We propose a methodology to
investigate coastal species biogeography and forecast changes in response to the
warming, able to account for the effects of other environmental stressors aside ocean
temperature.