Confronting collinearity: comparing methods for disentangling the effects of habitat loss and fragmentation

Estimating the relative importance of habitat loss and fragmentation is necessary to estimate the potential benefits of specific management actions and to ensure that limited conservation resources are used efficiently. However, estimating relative effects is complicated because the two processes are highly correlated. Previous studies have used a wide variety of statistical methods to separate their effects and we speculated that the published results may have been influenced by the methods used. We used simulations to determine whether, under identical conditions, the following 7 methods generate different estimates of relative importance for realistically correlated landscape predictors: residual regression, model or variable selection, averaged coefficients from all supported models, summed Akaike weights, classical variance partitioning, hierarchical variance partitioning, and a multiple regression model with no adjustments for collinearity. We found that different methods generated different rankings of the predictors and that some metrics were strongly biased. Residual regression and variance partitioning were highly biased by correlations among predictors and the bias depended on the direction of a predictor’s effect (positive vs. negative). Our results suggest that many efforts to deal with the correlation between amount and fragmentation may have done more harm than good. If confounding effects are controlled and adequate thought is given to the ecological mechanisms behind modeled predictors, then standardized partial regression coefficients are unbiased estimates of the relative importance of amount and fragmentation, even when predictors are highly correlated.     

The homogenizing influence of agriculture on forest bird communities at landscape scales

Landscape Ecology - Agricultural expansion is a principal driver of biodiversity loss, but the impacts on community assembly in agro-ecosystems are less clear, especially across regional scales at...     

Effect of road density on abundance of white-footed mice

While several studies have demonstrated that roads can act as barriers to small mammal movement, the relationship between road density and small mammal abundance has not yet been investigated. In southeastern Ontario, Peromyscus leucopus (white-footed mice) suffer high over-winter mortality rates, resulting in small springtime populations and frequent local extinctions. Peromyscus leucopus movement is known to be inhibited by roads, which should result in lower rates of immigration into and recolonization of habitats in landscapes with high road density. We tested two predictions: (1) Forest sites situated in landscapes with high road densities have a higher chance of P. leucopus being absent during the early spring than forest sites situated in landscapes with low road densities and (2) P. leucopus populations during the summer are smaller in forest sites situated in landscapes with high road densities than in landscapes with low road densities. We sampled P. leucopus in focal patches within nineteen landscapes (7 rural, low-road-density landscapes; 7 rural, high-road-density landscapes; 5 urban landscapes). There was no significant relationship between road density and the presence/absence of P. leucopus during the early spring. We found a significant positive effect of road density on P. leucopus relative abundance during the summer, even when we excluded the urban landscapes and based the analysis on only the 14 rural landscapes. Our results suggest that any negative effect of roads on P. leucopus populations, created by their inhibition to moving across roads, is far outweighed by some positive effect of roads on P. leucopus abundance. We suggest that the two most likely explanations are that roads are positively correlated with an important as-yet-undetermined component of habitat quality, or that roads positively affect P. leucopus by negatively affecting their predators.     

The borders within: Mobility and enclosure in the Riau Islands1

Abstract: The border studies literature makes a strong case against claims for unfettered transnationalism and ‘borderlessness’ in our ‘globalising world’. However, its focus on movement across borders means that it fails to address bordering practices that occur within the nation‐state as a result of transnational activity. In this paper, we extend Cunningham and Heyman’s concepts ‘enclosure’ and ‘mobility’ to confront the different layers of bordering (both physical and non‐physical) that have occurred in Indonesia’s Riau Islands since they became part of the Indonesia–Malaysia–Singapore Growth Triangle.     

Positive effects of forest fragmentation,independent of forest amount,on bat abundance in eastern Ontario,Canada

While studies have found that bat abundance is positively related to the amount of forest cover in a landscape, the effects of forest fragmentation (breaking apart of forest, independent of amount) are less certain, with some indirect evidence for positive effects of fragmentation. However, in most of these studies, the variables used to quantify fragmentation are confounded with forest amount, making it difficult to interpret the results. The purpose of this study was to examine how forest amount and forest fragmentation independently affect bat abundance. We conducted acoustic bat surveys at the centers of 22 landscapes throughout eastern Ontario, Canada, where landscapes were chosen to avoid a correlation between forest amount and forest fragmentation (number of patches) at multiple spatial scales, while simultaneously controlling for other variables that could affect bat activity. We found that the effects of forest amount on bat relative abundance were mixed across species (positive for Lasiurus borealis, negative for Perimyotis subflavus and Lasionycteris noctivagans). When there was evidence for an effect of forest fragmentation, independent of forest amount, on bat relative abundance, the effect was positive (Myotis septentrionalis, Myotis lucifugus and Lasiurus borealis). We suggest that the mechanism driving the positive responses to fragmentation is higher landscape complementation in more fragmented landscapes; that is, increased access to both foraging and roosting sites for these bat species. We conclude that fragmented landscapes that maximize complementation between roosting and foraging sites should support a higher diversity and abundance of bats.     

Predictors of pregnancy and postpartum haemoglobin concentrations in low-income women

OBJECTIVE: Pregnancy and postpartum iron status is of great public health importance, yet few studies have examined predictors of haemoglobin (Hb) concentration during this time. We identified predictors of Hb from 24 weeks' gestation until delivery and from 4 to 25 weeks postpartum. DESIGN: Blood was drawn as many as four times during care: at the initial visit, at 24-29 weeks' gestation, at delivery and postpartum. A longitudinal, multivariable linear regression model was used to predict Hb concentration. SETTING: A public health clinic in Raleigh, North Carolina.Subjects: n=520 women who participated in the Iron Supplementation Study. RESULTS: Hb concentration at the previous blood draw, short stature, non-Hispanic white ethnicity/race, >12 years of education and smoking were positive predictors of pregnancy and postpartum Hb concentrations. Iron supplement use was a positive predictor, while inadequate weight gain and severe nausea/vomiting were negative predictors of gestational Hb. A high infant birth weight and postpartum haemorrhage were negative predictors of postpartum Hb. Pre-pregnancy body mass index had a slight positive relationship with gestational Hb, but had a strong negative relationship with postpartum Hb. The longitudinal model also confirmed the typical pattern of gestational Hb concentration. As the number of weeks between the initial visit and the 24- to 29-week visit increased, Hb at 24-29 weeks' gestation decreased. As gestational age increased from 24 weeks until delivery, Hb concentration increased as well. CONCLUSIONS: The predictors identified here could be used in clinical settings to target high-risk women for intervention.     

Response of generalist and specialist insect herbivores to landscape spatial structure

The purpose of this study was to investigate the effect ofchanges in landscape pattern on generalist and specialistinsects. We did this by comparing the species richness andabundance of generalist and specialist herbivorous insects inalfalfa (Medicago sativa, L.) fields on 26 agriculturallandscapes that differed in spatial structure. The insects werefrom the families Curculionidae (Coleoptera), weevils, andCicadellidae (Auchennorhyncha), leafhoppers.We hypothesized that: (1) generalist richness and abundancewould be highest in landscapes with high diversity(Shannon-Wiener); (2) specialist richness and abundance would behighest in landscapes with (i) high percent cover alfalfa and(ii) low mean inter-patch distance. We tested for these effectsafter controlling for the patch-level effects of field size,field age, frequency of disturbance and vegetation texture.The important findings of the study are: (1) generalist richness andabundance increased with increasing landscape diversity and (2)isolation (percent cover alfalfa in the landscape and/or meaninter-patch distance) does not affect specialist insects. Theseresults are significant because they indicate that bothgeneralist and specialist insects may move over much largerdistances than previously thought. This is one of the firststudies to demonstrate a large scale effect of spatial structureon insects across a broad range of landscapes.     

What size is a biologically relevant landscape?

The spatial extent at which landscape structure best predicts population response, called the scale of effect, varies across species. An ability to predict the scale of effect of a landscape using species traits would make landscape study design more efficient and would enable landscape managers to plan at the appropriate scale. We used an individual based simulation model to predict how species traits influence the scale of effect. Specifically, we tested the effects of dispersal distance, reproductive rate, and informed movement behavior on the radius at which percent habitat cover best predicts population abundance in a focal area. Scale of effect for species with random movement behavior was compared to scale of effect for species with three (cumulative) levels of information use during dispersal: habitat based settlement, conspecific density based settlement, and gap-avoidance during movement. Consistent with a common belief among researchers, dispersal distance had a strong, positive influence on scale of effect. A general guideline for empiricists is to expect the radius of a landscape to be 4?C9 times the median dispersal distance or 0.3?C0.5 times the maximum dispersal distance of a species. Informed dispersal led to greater increases in population size than did increased reproductive rate. Similarly, informed dispersal led to more strongly decreased scales of effect than did reproductive rate. Most notably, gap-avoidance resulted in scales that were 0.2?C0.5 times those of non-avoidant species. This is the first study to generate testable hypotheses concerning the mechanisms underlying the scale at which populations respond to the landscape.