Modelling Laminaria hyperborea kelp forest distribution in ...ices.dk/sites/pub/CM...

1
T. Bekkby 1 *, E. Rinde 1 , K.M. Norderhaug 1 , H. Christie 1 , H. Gundersen 1 , L. Erikstad 2 and V. Bakkestuen 3 1 Norwegian Institute for Water Research, 2 Norwegian Institute for Nature Research, 3 University of Oslo *Contact author. Phone: +4722185100, fax: +4722185200, e-mail: [email protected] Modelling Laminaria hyperborea kelp forest distribution in Norway Laminaria hyperborea kelp forests cover large areas along the Norwe- gian coast, being highly productive, hosting a broad diversity of species and providing basis for commercial kelp harvesting and coastal fishing. Norway has a long and complex coastline, and detailed mapping is practically and economically difficult. Consequently, other approaches are needed. Statistical analyses, model selec- tion and spatial modelling was used to predict the distribution of L. hyperborea, identify kelp forest areas grazed by sea urchins (Stron- gylocentrotus droebachiensis) and predict the number of sea urchins within a region. The results show that geophysi- cal factors may predict the distribu- tion of kelp forest with a high degree of certainty, and that the distribution and number of seaurchins may be modelled. These results may for instance be used to estimate the magnitude of kelp forest loss and suggest areas for restoration initia- tives. Most relevant publications: • Bekkby, T., Rinde, E., Erikstad, L. and Bakkestuen, V. 2009. Spatial predictive distribution modelling of the kelp species Laminaria hyperborea. ICES Journal of Marine Science 66(10): 2106-2115. • Gundersen, H., Christie, H. and Rinde, E. 2010 (Norwegian, English summary). Sea urchins – from problem to commercial resource. Estimates of sea urchins as a re sources and an evaluation of ecological gains by sea urchin exploitation. NIVA report LNR 6001-2010. • Norderhaug, K.M., Isæus, M., Bekkby, T., Moy, F. and Pedersen, A. 2007. Spatial predictions of Laminaria hyperborea at the Norwegian Skagerrak coast. NIVA report LNR 5445-2007. + = and Gaustadalléen 21 • NO-0349 Oslo, Norway • Telephone: +47 22 18 51 00 • Fax: 22 18 52 00 • www.niva.no • [email protected] ICES CM 2010/Q:26 Density of sea urchins. The darker the blue, the higher the density. Hysværet Response Response 0.0 0.4 0.8 0.0 0.4 0.8 0.0 0.4 0.8 0.0 0.4 0.8 -40 -30 -20 -10 0 Depth (m) 0 5 10 15 Very sheltered Very exposed Wave exposure index -0.4 -0.5 0.0 0.2 0.4 0.6 Light exposure index Response Response 0 20 40 Terrain curvature (m) Probability Coverage 1 0,8 0,6 0,4 0,2 0 0 1 2 3 Field work - Field data were collected along geophysical gradients using underwater camera. Statistical analyses - We analysed the influence of geophysical factors using Generalised Additive Models (GAMs). As a tool for model selection, we used the Akaike Information Criterion (AIC). Predicting kelp forest distribution - Based on the response curves from the GAM analysis, we used the program GRASP to develop a matrix of predicted values (look-up tables) to produce the spatial probability model (AUC=0.79). The field observed classes of kelp density was positively correlated with predicted probabilities (F=51.98, p<0.001). Predicting sea urchin density - Areas of optimal sea urchin harvesting was modelled based on statistical results on how sea urchin densities varies with depth and latitude.

Transcript of Modelling Laminaria hyperborea kelp forest distribution in ...ices.dk/sites/pub/CM...

Page 1: Modelling Laminaria hyperborea kelp forest distribution in ...ices.dk/sites/pub/CM Doccuments/CM-2010/Q/Q2610.pdfTitle Modelling Laminaria hyperborea kelp forest distribution in Norway.

T. Bekkby1*, E. Rinde1, K.M. Norderhaug1, H. Christie1, H. Gundersen1, L. Erikstad2 and V. Bakkestuen3

1Norwegian Institute for Water Research, 2Norwegian Institute for Nature Research,

3University of Oslo

*Contact author. Phone: +4722185100, fax: +4722185200, e-mail: [email protected]

Modelling Laminaria hyperborea kelp forest distribution in Norway

Laminaria hyperborea kelp forests cover large areas along the Norwe-gian coast, being highly productive, hosting a broad diversity of species and providing basis for commercial kelp harvesting and coastal fishing.

Norway has a long and complex coastline, and detailed mapping is practically and economically difficult. Consequently, other approaches are needed.

Statistical analyses, model selec-tion and spatial modelling was used to predict the distribution of L. hyperborea, identify kelp forest areas grazed by sea urchins (Stron-gylocentrotus droebachiensis) and predict the number of sea urchins within a region.

The results show that geophysi-cal factors may predict the distribu-tion of kelp forest with a high degree of certainty, and that the distribution and number of seaurchins may be modelled. These results may for instance be used to estimate the magnitude of kelp forest loss and suggest areas for restoration initia-tives.

     

 

   

 

   

 

   

Most relevant publications:• Bekkby,T.,Rinde,E.,Erikstad,L.andBakkestuen,V.2009.SpatialpredictivedistributionmodellingofthekelpspeciesLaminaria hyperborea. ICES Journal of Marine Science 66(10): 2106-2115.• Gundersen,H.,Christie,H.andRinde,E.2010(Norwegian,Englishsummary).Seaurchins–fromproblemtocommercialresource.Estimatesofseaurchinsasare sources and an evaluation of ecological gains by sea urchin exploitation. NIVA report LNR 6001-2010.• Norderhaug,K.M.,Isæus,M.,Bekkby,T.,Moy,F.andPedersen,A.2007.SpatialpredictionsofLaminaria hyperborea at the Norwegian Skagerrak coast. NIVA report LNR 5445-2007.

+

=and

Gaustadalléen21•NO-0349Oslo,Norway•Telephone:+4722185100•Fax:22185200•www.niva.no•[email protected]

ICES CM 2010/Q:26

Density of sea urchins.The darker the blue, the higher the density.

Hysværet

Res

pons

eR

espo

nse

0.0

0.4

0.8

0.0

0.4

0.8

0.0

0.4

0.8

0.0

0.4

0.8

-40 -30 -20 -10 0

Depth (m)

0 5 10 15Very sheltered Very exposed Wave exposure index

-0.4 -0.5 0.0 0.2 0.4 0.6 Light exposure index

Res

pons

eR

espo

nse

0 20 40 Terrain curvature (m)

Prob

abili

ty

Coverage

1

0,8

0,6

0,4

0,2

00 1 2 3

Field work -Fielddatawerecollectedalonggeophysicalgradientsusingunderwatercamera.

Statistical analyses - We analysed the influence of geophysical factors using Generalised Additive Models (GAMs). As a tool for model selection, we used the Akaike Information Criterion (AIC).

Predicting kelp forest distribution - Based on the response curves from the GAM analysis, we used the program GRASPtodevelopamatrixofpredictedvalues(look-uptables)toproducethespatialprobabilitymodel(AUC=0.79).Thefieldobservedclassesofkelpdensitywaspositivelycorrelatedwithpredictedprobabilities(F=51.98,p<0.001).

Predicting sea urchin density - Areas of optimal sea urchin harvesting was modelled based on statistical results on how sea urchin densities varies with depth and latitude.