Julian Di Stefano

A confidence interval approach to data analysis

The objective of ecological experiments is often to determine whether observed effects are large enough to be ecologically important. Despite this, effect size measures and associated measures of precision are frequently missing in published ecological research. In many cases, P-values are the only information available with which to assess the ecological importance of observed effects, but they provide a poor means of assessment. It is argued that specifying an important effect size a priori and then presenting observed effects with their associated confidence intervals is often a more informative way of presenting ecological data. A hypothetical data set is analysed and interpreted using both P-values and confidence intervals and the results from these two approaches compared. Effects interpreted using P-values were either statistically significant or not, while confidence intervals provided information about statistical significance, the precision of the estimates, and produced a range within which values for the true effects might plausibly lie. The results show that both statistically significant (<0.001) and non-significant (0.100) P-values did not provide useful information about the importance of their associated effects. The capacity to use confidence intervals for analysing complex ANOVA designs is discussed and the implications of different data analysis and presentation techniques for forest management are considered.

Power analysis and sustainable forest management

This paper discusses power analysis in the context of sustainable forest management. It is suggested that a priori power analysis should be formally incorporated into the planning stage of all experiments designed to test whether forestry practices are sustainable. A priori power analysis enables researchers to estimate the probability of making a Type II error (i.e., finding no significant difference when one in fact exists). This information is critical in the statistical assessment of sustainable forestry, as unwittingly accepting a Type II error could result in poor management decisions. In addition, it is proposed that statistical assessments of sustainable forestry objectives can be more relevant if alpha (α) is liberated from its traditional value of 0.05. It is argued that in the context of sustainable forestry, making a Type II error can be more costly than making a Type I error. Consequently, it often makes sense for beta (β) to be small (say 0.05) and α to take on a larger value. In other situations the cost of making a Type I error may be more important, thus a procedure which enables researchers to determine a locally relevant α:β ratio is recommended.

Energy efficiency and the environment: the potential for energy efficient lighting to save energy and reduce carbon dioxide emissions at Melbourne University, Australia

In this study, the potential to improve the energy efficiency of lighting systems at Melbourne University was assessed. The cost effectiveness of different lighting technology alternatives was also calculated. Electricity used by existing 1.2 metre fluorescent lighting fixtures and four energy efficient lighting technology alternatives was compared. Relative to the existing system, installation of the four lighting technology alternatives would result in energy savings of 13.9%, 20.5%, 24.4% and 64.9%, respectively. If the technology alternative that saved the most electricity was installed, carbon dioxide emissions associated with the Universitys electricity use would be reduced by about 10%. Economic analysis shows that overall, none of the four technology alternatives are cost effective, although fine scale analysis shows that one technology alternative is cost effective in two out of the five room categories that were examined. Three barriers to the cost effective installation of energy efficient lighting technologies at Melbourne University are identified: (a) low lighting system operating hours, (b) the low cost of electricity and (c) the high cost of energy efficient lighting components.

Diet selection by the swamp wallaby (Wallabia bicolor): feeding strategies under conditions of changed food availability

Abstract Mammals can have either generalized (mixed) or specialized diets. We expected swamp wallabies (Wallabia bicolor) to have mixed diets, and predicted a negative relationship between the selection of a food type and its relative availability (negative frequency dependence). We collected data on diets in a native Eucalyptus forest where the relative availability of food types (defined as 5 plant functional groups: ferns, forbs, monocots, shrubs, and trees) had been altered by timber harvesting. A comparison of diets between individuals living in 2 habitat types (unharvested forest and 5-year-old regenerating areas) showed that in both habitats forbs were the major dietary component, although moderate amounts of shrubs and monocots also were consumed. Trees and ferns were eaten less at unharvested sites, and more at 5-year-old sites. Nonmetric multidimensional scaling followed by a multiresponse permutation procedure demonstrated a substantial difference in diet composition between the habitats (multiresponse permutation procedure: A = 0.20, P < 0.001), but when analyzed using an index of diet selection, the difference was smaller (A = 0.05, P = 0.04). Three alternative analyses demonstrated negative frequency dependence in many cases, a result generally consistent with a mixed feeding strategy. With the exception of tree foliage, selection was positively correlated with the relative availability of at least 1 other food type, and largely uncorrelated with 3 forage quality variables (nitrogen, water, and dry matter digestibility). Additional data at a finer resolution and in different seasons are required to test the generality of these conclusions.

River red gum (Eucalyptus camaldulensis): a review of ecosystem processes, seedling regeneration and silvicultural practice

Summary This review brings together disparate information about the red gum ecosystem. The climate and soils of the Murray Valley are described, and I outline the impact of river regulation, as well as a number of other human activities, on some of the plants and animals of the region. The review illustrates the fundamental interconnectedness of ecosystem processes, especially in riparian systems. The ecology of red gum regeneration and current silvicultural practices are examined, and the likely future of the red gum ecosystem is discussed. On balance, it would seem that the greatest public good lies in re-establishing and maintaining the natural values of the red gum ecosystem.

Detecting mammals in heterogeneous landscapes: implications for biodiversity monitoring and management

With terrestrial mammals facing worldwide declines there is an increasing need to effectively monitor populations so that appropriate conservation actions can be taken. There are many techniques available to survey terrestrial mammals and in recent years there have been a number of studies comparing the effectiveness of different methods. Most of these studies have not considered complementarity (the degree to which techniques detect unique species) and effectiveness across ecological gradients. In this study we examined three widely used techniques, camera trapping, live trapping and hair detection, for their complementarity across a vegetation and disturbance gradient. Overall, camera trapping detected more species than any other single technique, but live trapping complemented the cameras by consistently detecting unique species. Additionally, technique effectiveness differed between vegetation types; cameras alone were most effective in dry forest systems while cameras combined with live traps were most effective in wetter forest systems. These results suggest that care needs to be taken when sampling across heterogeneous landscapes because relying on one technique alone could result in certain taxa being systematically overlooked, leading to potentially erroneous conclusions.

Mammalian browsing damage in the Mt. Cole State forest, southeastern Australia: analysis of browsing patterns, spatial relationships and browse selection

Browsing by mammals can have a substantial impact on the survival, growth and form of regenerating forest trees. This study investigated the severity and distribution of mammalian browsing damage, and the degree to which eucalypt seedlings (Eucalyptus spp.) were selected for browsing on three regenerating native forest coupes in southeastern Australia. Damage was greatest 6 months after eucalypt germination and at this time, both messmate (Eucalyptus obliqua) and blue gum (E. globulus) seedlings were browsed at similar intensity. Nine and 12 months after germination, messmate was browsed more heavily than blue gum. Consistent relationships between browsing damage and both distance from the forest edge and wallaby density were not observed. Analysis of use and availability data for messmate, blue gum and manna gum (E. viminalis) seedlings showed that messmate was selected for browsing more than expected by chance while manna gum was browsed less than expected. Mammalian browsing affects a substantial portion of Victoria’s native forest estate, and could alter the species composition of future forests. Research into the ecology of browsing species and the development of non-lethal browsing reduction strategies is seen as a priority for the management of mammalian browsing in Victoria.Browsing by mammals can have a substantial impact on the survival, growth and form of regenerating forest trees. This study investigated the severity and distribution of mammalian browsing damage, and the degree to which eucalypt seedlings (Eucalyptus spp.) were selected for browsing on three regenerating native forest coupes in southeastern Australia. Damage was greatest 6 months after eucalypt germination and at this time, both messmate (Eucalyptus obliqua) and blue gum (E. globulus) seedlings were browsed at similar intensity. Nine and 12 months after germination, messmate was browsed more heavily than blue gum. Consistent relationships between browsing damage and both distance from the forest edge and wallaby density were not observed. Analysis of use and availability data for messmate, blue gum and manna gum (E. viminalis) seedlings showed that messmate was selected for browsing more than expected by chance while manna gum was browsed less than expected. Mammalian browsing affects a substantial portion of Victoria’s native forest estate, and could alter the species composition of future forests. Research into the ecology of browsing species and the development of non-lethal browsing reduction strategies is seen as a priority for the management of mammalian browsing in Victoria.

Responses of invasive predators and native prey to a prescribed forest fire

Fire shapes biome distribution and community composition worldwide, and is extensively used as a management tool in flammable landscapes. There is growing concern, however, that fire could increase the vulnerability of native fauna to invasive predators. We developed a conceptual model of the ways in which fire could influence predator–prey dynamics. Using a before–after, control–impact experiment, we then investigated the short-term effects of a prescribed fire on 2 globally significant invasive mesopredators (red fox, Vulpes vulpes, and feral cat, Felis catus) and their native mammalian prey in a fire-prone forest of southeastern Australia. We deployed motion-sensing cameras to assess species occurrence, collected predator scats to quantify diet and prey choice, and measured vegetation cover before and after fire. We examined the effects of the fire at the scale of the burn block (1,190 ha), and compared burned forest to unburned refuges. Pre-fire, invasive predators and large native herbivores were more likely to occur at sites with an open understory, whereas the occurrence of most small-and medium-sized native mammals was positively associated with understory cover. Fire reduced understory cover by more than 80%, and resulted in a 5-fold increase in the occurrence of invasive predators. Concurrently, relative consumption of medium-sized native mammals by foxes doubled, and selection of long-nosed bandicoots (Perameles nasuta) and short-beaked echidnas (Tachyglossus aculeatus) by foxes increased. Occurrence of bush rats (Rattus fuscipes) declined. It was unclear if fire also affected the occurrence of bandicoots or echidnas, as changes coincided with normal seasonal variations. Overall, prescribed fire promoted invasive predators, while disadvantaging their medium-sized native mammalian prey. Further replication and longer-term experiments are needed before these findings can be generalized. Nonetheless, such interactions could pose a serious threat to vulnerable species such as critical weight range mammals. Integrated invasive predator and fire management are recommended to improve biodiversity conservation in flammable ecosystems.

Bird functional diversity decreases with time since disturbance: Does patchy prescribed fire enhance ecosystem function?

Animal species diversity is often associated with time since disturbance, but the effects of disturbances such as fire on functional diversity are unknown. Functional diversity measures the range, abundance, and distribution of trait values in a community, and links changes in species composition with the consequences for ecosystem function. Improved understanding of the relationship between time since fire (TSF) and functional diversity is critical given that the frequency of both prescribed fire and wildfire is expected to increase. To address this knowledge gap, we examined responses of avian functional diversity to TSF and two direct measures of environmental heterogeneity, plant diversity, and structural heterogeneity. We surveyed birds across a 70-year chronosequence spanning four vegetation types in southeast Australia. Six bird functional traits were used to derive four functional diversity indices (richness, evenness, divergence, and dispersion) and the effects of TSF, plant diversity and structural heterogeneity on species richness and the functional diversity indices were examined using mixed models. We used a regression tree method to identify traits associated with species more common in young vegetation. Functional richness and dispersion were negatively associated with TSF in all vegetation types, suggesting that recent prescribed fire generates heterogeneous vegetation and provides greater opportunities for resource partitioning. Species richness was not significantly associated with TSF, and is probably an unreliable surrogate for functional diversity in fire-prone systems. A positive, relationship between functional evenness and structural heterogeneity was comnon to all vegetation types, suggesting that fine-scale (tens of meters) structural variation can enhance ecosystem function. Species more common in young vegetation were primarily linked by their specialist diets, indicating that ecosystem services such as seed dispersal and insect control are enhanced in more recently burnt vegetation. We suggest that patchy prescribed fire sustains functional diversity, and that controlled use of patchy fire to break up large expanses of mature vegetation will enhance ecosystem function.

Bird diversity increases after patchy prescribed fire: implications from a before-after control-impact study

Increasingly, patchy prescribed fire of low severity is used by land managers to mitigate wildfire risk, but there are relatively few experimental studies on the effects of low-severity fire on fauna. We used a before–after control–impact experiment to examine avian responses to prescribed fire at two scales in topographically variable, tall-open eucalypt forest in south-east Australia. We surveyed birds at control and impact areas twice before and twice after fire, and applied mixed models to investigate responses of avian turnover, richness and the occurrence of selected species. Approximately half of the impact area was burnt and topographic variation generated a finger-like configuration of burnt patches on ridges and unburnt patches in gullies. Our findings at the smaller scale (0.8 ha) indicated that the fire resulted in increased bird diversity because a patchwork of burnt and unburnt areas provided a mosaic of distinct successional states in which different species occurred. Additionally, we found that the effect of fire on species richness and occurrence was a function of the presence of unburnt topographic refuges. In contrast, we found no compelling evidence to suggest that birds responded to the fire at the larger scale (400 ha). We conclude that application of low-severity fire in a patchy manner enhanced avian diversity and facilitated the persistence of the birds detected in pre-fire surveys. Although the levels of patchiness required to sustain diverse taxa warrant further study, our findings highlight the importance of formally incorporating patchiness into prescribed burning for the ecologically sensitive management of contemporary landscapes.