@article{GrimmRevillaGroeneveldetal.2005,
  author    = {Grimm, Volker and Revilla, Eloy and Groeneveld, J{\"u}rgen and Kramer-Schadt, Stephanie and Schwager, Monika and Tews, J{\"o}rg and Wichmann, Matthias and Jeltsch, Florian},
  title     = {Importance of buffer mechanisms for population viability analysis},
  year      = {2005},
  language  = {en}
}
@phdthesis{Schwager2005,
  author    = {Schwager, Monika},
  title     = {Climate change, variable colony sizes and temporal autocorrelation : consequences of living in changing environments},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus-5744},
  school      = {Universit{\"a}t Potsdam},
  year      = {2005},
  abstract  = {Natural and human induced environmental changes affect populations at different time scales. If they occur in a spatial heterogeneous way, they cause spatial variation in abundance. In this thesis I addressed three topics, all related to the question, how environmental changes influence population dynamics. In the first part, I analysed the effect of positive temporal autocorrelation in environmental noise on the extinction risk of a population, using a simple population model. The effect of autocorrelation depended on the magnitude of the effect of single catastrophic events of bad environmental conditions on a population. If a population was threatened by extinction only, when bad conditions occurred repeatedly, positive autocorrelation increased extinction risk. If a population could become extinct, even if bad conditions occurred only once, positive autocorrelation decreased extinction risk. These opposing effects could be explained by two features of an autocorrelated time series. On the one hand, positive autocorrelation increased the probability of series of bad environmental conditions, implying a negative effect on populations. On the other hand, aggregation of bad years also implied longer periods with relatively good conditions. Therefore, for a given time period, the overall probability of occurrence of at least one extremely bad year was reduced in autocorrelated noise. This can imply a positive effect on populations. The results could solve a contradiction in the literature, where opposing effects of autocorrelated noise were found in very similar population models. In the second part, I compared two approaches, which are commonly used for predicting effects of climate change on future abundance and distribution of species: a "space for time approach", where predictions are based on the geographic pattern of current abundance in relation to climate, and a "population modelling approach" which is based on correlations between demographic parameters and the inter-annual variation of climate. In this case study, I compared the two approaches for predicting the effect of a shift in mean precipitation on a population of the sociable weaver Philetairus socius, a common colonially living passerine bird of semiarid savannahs of southern Africa. In the space for time approach, I compared abundance and population structure of the sociable weaver in two areas with highly different mean annual precipitation. The analysis showed no difference between the two populations. This result, as well as the wide distribution range of the species, would lead to the prediction of no sensitive response of the species to a slight shift in mean precipitation. In contrast, the population modelling approach, based on a correlation between reproductive success and rainfall, predicted a sensitive response in most model types. The inconsistency of predictions was confirmed in a cross-validation between the two approaches. I concluded that the inconsistency was caused, because the two approaches reflect different time scales. On a short time scale, the population may respond sensitively to rainfall. However, on a long time scale, or in a regional comparison, the response may be compensated or buffered by a variety of mechanisms. These may include behavioural or life history adaptations, shifts in the interactions with other species, or differences in the physical environment. The study implies that understanding, how such mechanisms work, and at what time scale they would follow climate change, is a crucial precondition for predicting ecological consequences of climate change. In the third part of the thesis, I tested why colony sizes of the sociable weaver are highly variable. The high variation of colony sizes is surprising, as in studies on coloniality it is often assumed that an optimal colony size exists, in which individual bird fitness is maximized. Following this assumption, the pattern of bird dispersal should keep colony sizes near an optimum. However, I showed by analysing data on reproductive success and survival that for the sociable weaver fitness in relation to colony size did not follow an optimum curve. Instead, positive and negative effects of living in large colonies overlaid each other in a way that fitness was generally close to one, and density dependence was low. I showed in a population model, which included an evolutionary optimisation process of dispersal that this specific shape of the fitness function could lead to a dispersal strategy, where the variation of colony sizes was maintained.},
  subject      = {Populationsbiologie},
  language  = {en}
}
@article{WichmannJohstSchwageretal.2005,
  author    = {Wichmann, Matthias and Johst, Karin and Schwager, Monika and Jeltsch, Florian and Blasius, Bernd},
  title     = {Extinction risk, coloured noise and the scaling of variance},
  year      = {2005},
  abstract  = {The impact of temporally correlated fluctuating environments (coloured noise) on the extinction risk of populations has become a main focus in theoretical population ecology. In this study we particularly focus on the extinction risk in strongly autocorrelated environments. Here, in contrast to moderate autocorrelation, we found the extinction risk to be highly dependent on the process of noise generation, in particular on the method of variance scaling. Such variance scaling is commonly applied to avoid variance-driven biases when comparing the extinction risk for white and coloured noise. In this study we found an often-used scaling technique to lead to high variability in the resulting variances of different time series for strong auto-correlation eventually leading to deviations in the projected extinction risk. Therefore, we present an alternative method that always delivers the target variance, even in the case of strong temporal correlation. Furthermore, in contrast to the earlier method, our very intuitive method is not bound to auto-regressive processes but can be applied to all types of coloured noises. We recommend the method introduced here to be used when the target of interest is the effect of noise colour on extinction risk not obscured by any variance effects.},
  language  = {en}
}