Managing Monarch butterfly overwintering groves: making room among the eucalyptus Proper management and conservation of the coastal California overwintering sites used by western Monarch butterflies (Danaus plexippus L.) is critical for continued use of these sites by monarchs. Many management efforts are currently concentrating on eucalyptus-only sites because of the prevailing notion that monarchs prefer eucalyptus over native tree species. Yet, whether a preference exists or not has never been tested. Herein, we test the “eucalyptus preference” hypothesis with data from five overwintering sites comprised of blue gum eucalyptus (Eucalyptus globulus) and at least one other native tree species from fall 2009 to spring 2012. We found that when monarchs clustered disproportionately on a tree species relative to its availability, they clustered significantly more than expected on native trees and significantly less than expected on eucalyptus. Also, in years when the overwintering population was highest, monarchs clustered disproportionately on native conifers, and they often switched from clustering on eucalyptus in the early winter to native conifers in the middle or late winter. Our results suggest that overwintering groves should be managed to include a mixture of tree species. We cannot recommend simply planting more eucalyptus. At overwintering sites in central coastal California, native conifers such as Monterey cypress (Hesperocyparis macrocarpa) and pitch canker-resistant Monterey pine (Pinus radiata) should be planted as replacements for blue gum eucalyptus in areas where trees are likely to fall, and around the perimeter of groves. Testing the Monarch butterfly eucalyptus preference hypothesis at California overwintering sites Western Monarch butterflies (Danaus plexippus L.) overwinter in groves of native and non-native trees along the California coast. Eucalyptus is abundant in coastal counties, and overwintering monarchs utilize this type of tree more than any other. This has led to the belief that monarchs prefer eucalyptus. Yet whether a preference exists has never been tested. We tested the “eucalyptus preference” hypothesis at five California overwintering sites with canopies comprised of eucalyptus and at least one native conifer species. We found that at no time over the course of three years did monarchs cluster on trees in proportion to their availability in the canopy. Overall, they did not cluster on one tree species significantly more frequently than another, indicating that monarchs do not prefer eucalyptus—or any tree species—all of the time. However, more often than not monarchs clustered significantly more than expected on native trees, particularly at midseason when the weather was most inclement. They also clustered disproportionately on native conifers when the overwintering population size was highest. At most sites monarchs exhibited tree switching, shifting from eucalyptus to native conifers in the middle or late winter. Based on these results, we reject the “eucalyptus preference” hypothesis. In its place, we propose the “conditional preference hypothesis”, wherein monarchs are predicted to prefer cluster trees according to microclimate conditions and prefer alternate trees within a site as climatic conditions change. Rejection of the eucalyptus preference hypothesis suggests that sites comprised exclusively of eucalyptus may not offer monarchs a suitable range of microhabitats, and further suggests we should rethink “eucalyptus-centric” management. Monarch butterflies overwintering in coastal California: low site fidelity and high intersite movement Western monarch butterflies (Danaus plexippus L.) overwinter in large aggregations at hundreds of sites along the California coast. Management plans and census methods are both founded on the assumption that individual monarchs arrive at an overwintering site in the fall and stay at that site for the winter. Though populations potentially coalesce en masse from autumnal sites onto climax overwintering sites, very little individual movement between sites is inferred. Monarch movement is therefore thought to be primarily into sites (as opposed to out of or among them). We refer to this assumption and inference as the accrual hypothesis. In light of previous studies that provide evidence for movement among sites, we propose that overwintering monarchs may belong to a superpopulation. The existence of a superpopulation comprised of individuals moving in and out of sites would force us to rethink our ideas of landscape-level resource use by monarchs, our site-centric (rather than landscape-level) management strategies, and our abundance estimation techniques, which employ closed population models. We tested the closed population model, the accrual hypothesis, and the superpopulation model at three California overwintering sites using a mark-resight study design. We found that a large proportion of the monarchs at a site moved among (into and out of) monitored sites, both while the population size increased in October and November, and while the population appeared to exhibit an equilibrium winter maximum. The pattern of abundance of both tagged and untagged monarchs at monitored sites leads us to reject the closed population model and the accrual hypothesis. We found that monarchs at all three study sites are part of a larger superpopulation, though the sites do not contribute to the superpopulation equally. We determined that mark-resight is a viable alternative to existing population estimation techniques, though mark-resight methods would need to be explored further before being applied routinely. Our results suggest we need to move away from site-based management and manage instead for landscape-level overwintering (superpopulation) dynamics.

Étant donné un modèle de super-population (des variables aléatoires sont générées indépendamment et selon une même loi initiale sur une population) et un plan de sondage informatif, une loi de probabilité limite et une densité de probabilité limite des observations sur l'échantillon sont définies correspondant à des tailles de population et d'échantillon tendant vers l'infini. Le processus aléatoire de sélection peut induire une dépendance entre les observations sélectionnés. Un cadre asymptotique et des conditions faibles sur le processus de sélection sont donnés, sous lesquels les propriétés asymptotiques classiques sont conservées malgré la dépendance des données : la convergence uniforme de la fonction de répartition empirique. Par ailleurs, nous donnons la vitesse de convergence de l'estimateur à noyau de la densité vers la densité limite de l'échantillon. Ces résultats constituent des indications selon lesquelles il est parfois possible de considérer que les réalisations sur l'échantillon sont id et suivent approximativement la densité limite définie, notamment dans une perspective d'inférence sur le modèle de super-population. Par exemple, étant donné un modèle paramétrique on peut définir la vraisemblance approchée de l'échantillon comme produit de densités limites et un estimateur de maximum de vraisemblance approchée, dont on établit la normalité asymptotique . La dernière partie traite de tirage équilibré : des algorithmes de calcul de probabilités d'inclusion minimisant une approximation de la variance de l'estimateur de Horvitz-Thompson d'un total sont proposés.