All evaluation of a large, operational full-text document-retrieval system (containing roughly 350,000 pages of text) shows the system to be retrieving less than 20 percent of the documents relevant to a particular search. The findings are discussed in terms of the theory and practice of full-text document retrieval. Marchl985Volume 28 Number 3 Cotnmunications of the ACM

Large old trees are critical organisms and ecological structures in forests, woodlands, savannas, and agricultural and urban environments. They play many essential ecological roles ranging from the storage of large amounts of carbon to the provision of key habitats for wildlife. Some of these roles cannot be replaced by other structures. Large old trees are disproportionately vulnerable to loss in many ecosystems worldwide as a result of accelerated rates of mortality, impaired recruitment, or both. Drivers of loss, such as the combined impacts of fire and browsing by domestic or native herbivores, chemical spray drift in agricultural environments, and postdisturbance salvage logging, are often unique to large old trees but also represent ecosystem-specific threats. Here, we argue that new policies and practices are urgently needed to conserve existing large old trees and restore ecologically effective and viable populations of such trees by managing trees and forests on much longer time scales than is currently practiced, and by protecting places where they are most likely to develop. Without these steps, large old trees will vanish from many ecosystems, and associated biota and ecosystem functions will be severely diminished or lost.

Understanding the impacts of natural and human disturbances on forest biota is critical for improving forest management. Many studies have examined the separate impacts on fauna and flora of wildfire, conventional logging, and salvage logging, but empirical comparisons across a broad gradient of simultaneous disturbances are lacking. We quantified species richness and frequency of occurrence of vascular plants, and functional group responses, across a gradient of disturbances that occurred concurrently in 2009 in the mountain ash forests of southeastern Australia. Our study encompassed replicated sites in undisturbed forest (~70 yr post fire), forest burned at low severity, forest burned at high severity, unburned forest that was clearcut logged, and forest burned at high severity that was clearcut salvage logged post-fire. All sites were sampled 2 and 3 yr post fire. Mean species richness decreased across the disturbance gradient from 30.1 species/site on low-severity burned sites and 28.9 species/site on high-severity burned sites, to 25.1 species/site on clearcut sites and 21.7 species/site on salvage logged sites. Low-severity burned sites were significantly more species-rich than clearcut sites and salvage logged sites; high-severity burned sites supported greater species richness than salvage logged sites. Specific traits influenced species' sensitivity to disturbance. Resprouting species dominated undisturbed mountain ash forests, but declined significantly across the gradient. Fern and midstory trees decreased significantly in frequency of occurrence across the gradient. Ferns (excluding bracken) decreased from 34% of plants in undisturbed forest to 3% on salvage logged sites. High-severity burned sites supported a greater frequency of occurrence and species richness of midstory trees compared to clearcut and salvage logged sites. Salvage logging supported fewer midstory trees than any other disturbance category, and were distinctly different from clearcut sites. Plant life form groups, including midstory trees, shrubs, and ferns, were dominated by very few species on logged sites. The differences in biotic response across the gradient of natural and human disturbances have significant management implications, particularly the need to reduce mechanical disturbance overall and to leave specific areas with no mechanical disturbance across the cut area during logging operations, to ensure the persistence of resprouting taxa.

We applied an ecosystem risk assessment to the mountain ash forest ecosystem of the Central Highlands of Victoria (hereafter 'mountain ash forest'), south-eastern Australia, using the IUCN Red List of Ecosystems criteria. Using this methodology, we quantified: (i) key aspects of the ecosystem's historical, current and future decline in spatial distribution; (ii) the extent of occurrence and area of occupancy for the mountain ash ecosystem; and (iii) the decline in key abiotic and biotic processes and features for historical, current and future time periods. We developed a probabilistic model of tree growth stages to estimate the risk of ecosystem collapse within 50 to 100 years in the mountain ash forest. There was uncertainty in our estimates of risk under the various IUCN criteria, with two sub-criteria being categorized as 'Data Deficient'. Our overall ranking of risk of collapse for the ecosystem was Critically Endangered. We are confident that this risk category is appropriate because all 39 scenarios modelled indicated a ≥92% chance of ecosystem collapse by 2067. Our findings highlight the important need for timely policy reform to facilitate improved management of the mountain ash ecosystem in Victoria. In particular, there needs to be greater protection of remaining areas of unburned forest, and restoration activities in parts of the forest estate. Implementation of these strategies will require a significant reduction in logging pressure on the mountain ash ecosystem.