Data issues in species monitoring: where are the traps? French breeding bird monitoring : Animations locales : - dealing with heterogenous data - working with multi-species multi-sites monitoring schemes - from monitoring data to biodiversity indicators - learning effect - believe there is nothing in there Sampling design 7 carrés suivis au moins une fois entre et 6 5 espèces Publication from the French Breeding Bird Monitoring scheme -7 Ecology Letter 6, 6 Global Change Biology, 7 Prodeedings Royal Society B Conservation Biology, 7 Oikos 7 Global Ecology and Biogeography 7 Biodiversity & Conservation 5, 6, 6 Agriculture Ecosystem & Environment 6 Biological Conservation 7 Bird Study 5 N = Et avec le soutien de
Impliquer le grand public dans le suivi de la biodiversité Un réseau national 8 espèces de papillons communs Suivi scientifique 695 personnes inscrites et abonnées à la lettre d information 5 jardins en 6 Eveil de l intérêt naturaliste 98 jardins en 7 5 papillons comptés Changement de comportement Cartes d abondance relative phénology data Évaluation des «vrong positifs» (< 5%)
Objectif scientifique Evaluer l influence du paysage sur la diversité en papillons Indice d abondance 6 5...6.8 % urbanisation % de milieu artificiel dans la commune 6 espèces / 8 Missing points are not a problem General case: volunteers cannot take measures at some sites in some years Difficulty: site and observer effects make statistical units not interchangeable Indice d abondance.5..7..9.5...6.8 % de milieu artificiel dans la commune % urbanisation Solution: standardization for site effect = accounting for site effect in model Thus no need for continuity of time series
Nombre d'individus comptés standardisé par site Nombre d'individus comptés Acounting for missing data & heterogeneity 6 - - - -5 8 6 Site A Site B 988 989 99 99 99 99 99 995 996 site A Site B Variation moyenne 988 989 99 99 99 99 99 995 996 - Exemple : Skylark Indice d'abondance 9 8 7 988 99 99 997 Pool sites -.7 ±.6 per year exponentiel (-.7 * ) - - 8% between 989 et....9.8.7.6 Pool species Indicateur toutes espèces : -7 % d oiseaux entre 989 et 5 Indicateurs par milieux : Généralistes +7% 989 99 99 995 997 999 5 Urban % All species - 7% Woodland - 7% Farmland - 9% Indicator of farmland birds in Europe Pooling countries Population index (99=) 5 9 7 Significant decline (trend 98- = -9%) 978 98 986 99 99 998 Year Gregory et al. Phil. Trans. R. Soc. Lond. B. (in press).
Pooling, pooling, pooling We know how to do that, But if we stop there, we have mostly destroyed data: there is much to learn from the differences among sites, species etc. Ex. winners and losers in fragmented lanscape For each species : abondance(*) ~ fragmentation abondance(*) ~ perturbation Abondance? x species perturbation Species ranked according to habitat specialisation (*) adjusted for habitat Estimation de la spécialisation Descripteur du paysage () CORINE landcover (99) Fragmentation Relative abundance 5 6 7 8 9 5 6 7 8 Relative abundance 5 6 7 8 9 5 6 7 8 Habitat classes Habitat classes Km Indice de spécialisation = CV(abondance)
Résults Specialized species do not like fragmentation Generalist species are more resiliant to fragmentation Most generalist species actualy enjoy fragmentation Slope abundance ~ fragmentation + R²=%.5.5.5 - Développer des indicateurs régionaux Species specialisation From monitoring data to biodiversity indicators From monitoring data to biodiversity indicators Strong demand for biodiversiy indicators => monitoring requested Trap n Strong demand for biodiversiy indicators => monitoring requested Monitoring Indicators Monitoring Indicators Data base for analysis Basic and applied ecological knowledge
From monitoring data to biodiversity indicators French breeding bird monitoring : Animations locales : Strong demand for biodiversiy indicators Monitoring => monitoring requested Indicators Trap n Sampling design 7 carrés suivis au moins une fois entre et 6 5 espèces Learning effect in the FBBS Started in, new participants every years From, we compare changes between first and second year squares, with changes in «older» squares => Observers tend to count % more individuals after the first year, whatever the species Learning effect in the FBBS Linked to random choice + point counts? Solution : remove first year Solution : use appropriate stats