This study reports the population responses of the lizard Psammodromus algirus to forest fragmentation by comparing its distribution in two contrasting situations of habitat conservation: a well connected mosaic landscape (forest patches larger than 2000 ha and/or connected by corridors that prevent their isolation), and an archipelago of forest remnants embedded within a matrix of cereal fields. The frequency of occurrence of P. algirus was larger in the unfragmented habitats (14 out of 19 censused plots) than in the fragments (two out of of 21 wood-lots). Vegetation structure was a good predictor of lizard occurrence under conditions of no fragmentation, but not in the fragments, where high plant cover seemed a necessary, but insufficient, condition for the survival of lizard populations, and where remnant size was the only variable that differed significantly between wood-lots with and without lizards. Historical fragmentation of the habitat is also crucial for understanding the current distribution of lizards, as shown by their absence from a large forest that was highly fragmented in the past but which has been regenerating for decades. It is hypothesized that the combined effects of fragmentation and predation in small remnants have led to the extinction of P. algirus in fragments smaller than c. 90 ha, recolonization being prevented by the very limited dispersal abilities of these forest lizards.

We reintroduced 33 peregrine falcons (Falco peregrinus) at two release areas with contrasting habitat configurations to assess landscape influences on dispersal. One site (Daniel Boone National Forest) had non-forested corridors within a forest matrix, whereas the other site (Tom Dorman State Nature Preserve) had forested corridors within an agricultural matrix. We used aerial telemetry and ground observation to assess differences in dispersal between sites and we used multivariate statistical analyses to detect combinations of response variables depicting landscape influences. Non-forested corridors at Daniel Boone affected post-fledging movements and initiation of dispersal, whereas no discernable pattern in dispersal initiation was observed at Dorman Preserve. Results included significant (p<0.05) differences between peregrines at Daniel Boone and Dorman Preserve in $\bar{x}{\pm}$ SD post-fledging area (PFA) size (2643.5±2599.0 and 931.7±732.1 ha, respectively), maximum movement distances (12.9±13.9 and 6.1±4.6 km, respectively), orientation of movements (second-order $\bar{x}$ angle±95% confidence interval (CI)=280.8±58.6 (non-random orientation) and 358.8±98.8 (random orientation), respectively), time spent on the PFA (16.0±12.2 and 31.0±3.3 days, respectively) and selection of agricultural habitats (Euclidean distance vectors ρ=0.66 and ρ=2.50, respectively). Selection of agriculture by peregrines released on Daniel Boone reflected selection of corridors as part of pre-dispersal movement. These results, coupled with generally unorientated movement behaviour by Dorman Preserve peregrines, provided evidence for an influence of functional landscape connectivity on dispersal in peregrines.

Despite the alarming rate of tropical deforestation, the long-term conservation value of forest fragments remains poorly understood. We report on the avifaunal turnover in an isolated 4 ha tropical forest fragment in Singapore (i.e. Singapore Botanic Gardens rainforest fragment (SBGRF)) between 1898 and 1998. Over 100 years, the SBGRF lost 18 (49%) species and gained 20 species. More forest-dependent species (3) were lost from the SBGRF than survived (1) or colonised it (no species). Conversely, significantly more introduced species (4) colonised the fragment than were previously recorded (1 species). Significantly more nectarivores survived (8 species) or colonised (9 species) than were lost (two species). In essence, while the avian species richness in the SBGRF remained relatively constant after a century, its species composition underwent significant changes. The avian species composition in the SBGRF in 1998 appeared to be more similar to that of the contemporary smaller and younger Singaporean secondary forest patches than to either the larger and older forest reserves or to the SBGRF 100 years ago. Our study suggests that small isolated tropical forest fragments may have limited long-term conservation value for native forest bird species.

Anecdotal data gathered from many populations suggest that southern flying squirrel (SFS, Glaucomys volans) use of the endangered red-cockaded woodpecker's (RCW, Picoides borealis) nest and roost cavities may negatively affect RCW populations. We conducted a controlled experiment to determine the effects of SFSs on RCW reproductive success. During the 1994 and 1995 breeding seasons, SFSs were removed from 30 RCW clusters and 32 clusters served as controls. SFSs were the most frequently encountered occupants of RCW cavities and used 20–33% of RCW cavities in control and treatment clusters over both years. Treatment groups produced significantly more successful nests (≥ 1 fledgling) than control groups in 1994. In 1995 however, there was no difference in the number of successful nests. In both years, RCW groups nesting in treatment clusters produced significantly more fledglings than groups in control clusters in each of four experimental areas, averaging approximately 0.7 additional fledglings per nesting group. Loss of entire clutches or broods, possibly as a result of predation or abandonment, was a major factor limiting reproduction in control groups in 1994. In contrast, differences in partial brood loss appeared to be the cause of differential fledging success in 1995. Usurpation of RCW roost cavities by SFSs may have placed greater energetic demands on RCWs for cavity defence or thermoregulation, thus reducing energy available for reproduction. Our results show that SFS use of RCW cavities during the breeding season has a significant impact on RCWs and that management of RCW populations should include activities that either minimize SFS populations in RCW clusters or limit access of SFSs to RCW cavities.

The amount of genetic differentiation between the African wild cat (Felis lybica) and the domestic cat (F. catus) in southern Africa is not known, but there is concern that interbreeding between the two may have been sufficiently extensive to threaten the status of the wild cat as a genetically distinct population. Eight hypervariable microsatellite loci were therefore used to determine the genetic relationships within and among wild and domestic cat populations. Measures of genetic differentiation, including FST and RST, confirmed significant differentiation between the wild and domestic cat populations and no significant differentiation between two widely separate domestic cat populations. Our data indicate that the level of genetic introgression is lower than previously thought. This enhances the conservation status of the African wild cat and demonstrates the need for the implementation of feral cat control in and around game reserves and rural areas.

Inbreeding depression is a critical phenomenon for conservation of small populations, especially concerning the captive breeding of endangered animals whose effective population size may be very low. In order to manage such captive populations on a long-term basis, assessment of the interactions between demographic and genetic parameters and their consequences on the population survival probability is essential. Therefore, we examined (1) the relative importance of initial population size and carrying capacity on the survival probability of a population, (2) the impact of a bottleneck in an already inbred population. For this purpose, a stochastic model taking into account the effect of genes responsible for inbreeding depression was built. It was concluded that the population survival probability is very sensitive to the carrying capacity. Moreover, in inbred populations, a large increase in the extinction rate results from a second bottleneck. With two founders, survival is high as long as the carrying capacity is not very small (above a few decades), whereas a second bottleneck has a dramatic effect, even when the carrying capacity approaches 100. Consequently, previous inbreeding should be considered in the case of reintroduction. Reason for reintroduction failure could be successive bottlenecks, with the associated deleterious effects of inbreeding.

Habitat selection by translocated black howler monkeys (Alouatta pigra) was studied in Belize, Central America. Ranging patterns of two recently translocated groups were contrasted with those of two groups in the same area with established home ranges, on a yearly and monthly basis. All groups concentrated their activities along stream beds at elevations below 200 m. Newly translocated groups increased the percentage of their monthly ranges in riverine areas (within 100 m of a water source) over the year of the study (with the exception of the last month) while established groups did not. Areas used more than five times by the monkeys contained larger trees and a greater relative coverage of major food species than low-use areas even though they had a lower overall species richness and diversity. These monkeys may be selecting habitat within the forest based upon vegetative differences and ranging patterns may be affected by patterns of food availability.

Bushmeat markets can reflect the impact of hunting in large geographical areas, but efficient and adequate sampling strategies are needed. Five bushmeat markets in West and Central Africa were used to simulate the performance of different sampling regimes. The studied markets (n = 863 days) varied in animal carcasses and number of species recorded. In our simulations, we varied number of days sampled and their temporal distribution using a variant of the Monte Carlo methodology. Three sample strategies were considered: (1) unconstrained random sampling; (2) random sampling of start-points, where the n days sampled are a sequential block following the randomly selected start-point and (3) sampling blocked by season. No substantial differences between standard sampling theory and our simulations were present. However, only a large sample of markets will allow useful inferences on a regional scale and timing and coordination of sampling may be highly influential. Sampling in blocks of days was as efficient as simple random sampling in estimating species richness, but not carcass volume. This may indicate that, even with seasonality in market compositions, or irregular influences, the temporal pattern as described by presence/absence varies much less than does the volume of carcasses. Shorter sampling periods perform poorly in estimating species richness. The relationship between % species richness and overall carcass volume may predict the sampling effort required to estimate market species richness based on volume, when a large enough sample of markets becomes available. Similarly, a larger sample of markets would reveal how far the species composition in markets reflects the general organising principles of community structure in terms of frequency and abundance relationships.

The Przewalski's gazelle (Procapra przewalskii) is critically endangered and has experienced significant decline in population numbers in the past, and now survives in four isolated populations: the Bird Island, Hudong-Ketu, Yuanzhe and Shadao-Gahai populations. To understand better the effect of habitat fragmentation and geographical isolation on gene flow and genetic variation, and uncover genetic units for conservation, we examined the hypervariable region (420 bp) of the mitochondrial DNA control region of 29 Przewalski's gazelles from four localities distributed throughout their present geographical range. Phylogenetic analysis consistently revealed four distinct mtDNA lineages within Przewalski's gazelle, with each lineage restricted to a single geographical location. The study revealed a marked divergence of lineages from the Bird Island population relative to other populations, which corresponds to the geographical association. The nucleotide diversity within populations was very low (less than 0.004). The value of Nm was less than 0.3 individuals per generation, indicating low gene flow between populations. It was suggested that the Yuanzhe, Hudong-Ketu and Shadao-Gahai populations be managed as a conservation unit because of their genetic and geographical closeness.

Habitat clearance generally produces fragmented landscapes. A major consequence is the creation of large amounts of new ‘edge habitat’, often of a kind not previously existing, such as abrupt forest- agricultural land interfaces. Much work from around the world suggests that proliferation of edge habitat seriously affects birds reliant on large forest blocks to persist, and the intrusion of edge specialists into areas they did not previously occupy. These processes often generate avifaunal gradients in which assemblages change from the interior to the edge. This effect is explored in the largest remnant block of box-ironbark forest of central Victoria, Australia (about 30,000 ha). Eight radially oriented survey lines were established around the periphery of the block. Along each survey line, five transects with long axes oriented parallel to the edge were positioned with midlines at 40 m, 160 m, 280 m and 400 m into the forest, and one transect 80 m into the agricultural land, yielding a total of 40 transects. Six identical transects were located deep within the forest (> 2 km). There was little evidence of a change in the avifauna from interior to edge, although mean richness was depressed in the edge habitats compared with forest interior transects. These results were essentially the same for ‘all species’ or ‘forest dependent’ species. Thus, there is little evidence for marked edge effects on the avifauna of these large forest blocks, although work in intermediate sized remnants (100-1000 ha) is needed to identify thresholds at which edge effects begin to be manifested.

Genetic variation at the major histocompatibility (MHC) class I loci was examined in captive and wild samples of three subspecies of tiger (Bengal, Panthera tigris tigris; Siberian, P. t. altaica; Sumatran, P. t. sumatrae) by Southern blot analysis using the domestic cat probe FLA 24. All captive populations showed at least moderate variability compared to other felids. Captive tigers were, on average, significantly more variable than wild tigers, but this was not consistent among subspecies. Significant geographic genetic population structure was found at the MHC class I loci among both captive and wild tigers. Phylogenetic analysis of both captive and wild tigers placed Sumatran tigers basally. In a combined phylogenetic analysis, captive and wild tigers of the same subspecies occurred near one another, indicating that the structure of the captive populations reflects that of the wild populations.

We sequenced part of the mitochondrial DNA control region of the extinct Arabian ostrich, Struthio camelus syriacus, to determine its phylogenetic relationship to the surviving subspecies, and to provide genetic information pertinent to the reintroduction of contemporary ostrich into Saudi Arabia. Parsimony, neighbour joining, maximum likelihood and a minimum spanning network all supported a close genetic association between the Arabian S. c. syriacus and the north African S. c. camelus. Furthermore, all our analyses suggest a recent common ancestry for the southern African S. c. australis and the east African S. c. massaicus. The Somali race, S. c. molybdophanes, was phylogenetically the most distinct of the ostrich taxa. Our data are supportive of a management decision to introduce S. c. camelus into areas once occupied by the extinct S. c. syriacus based predominantly on their geographic proximity and phenotypic features. Moreover, the presence of a shared lineage in these taxa indicates that gene flow between the two geographic forms may have been possible in the recent evolutionary past, probably along the Egyptian–Sinai–Israel passageway.

As the human population increases, wildlife is becoming restricted to protected remnants of ecosystems with boundaries that limit dispersal. The effect on large herbivores generally is an increase in population density. We investigated the relationship between population density and reproduction in white rhinos (Ceratotherium simum) in Matobo National Park, Zimbabwe. Two lines of evidence were examined: (1) a contrast in reproduction between a high-density and a low-density population; and (2) a longitudinal analysis of changes in reproductive parameters of a fenced population as density increased by >200%. As density increased within the fenced area, the rates of population growth and recruitment of calves decreased, and the age at first reproduction for females increased significantly. Females in the low-density population produced their first calves at significantly younger ages than those in the high-density population. Loss of body condition was related to both population density and female reproductive status. Because most African rhinos exist in relatively small reserves, an understanding of the implications of restricted dispersal and increased population density on their demography is critical to their conservation in situ. Managers need to consider trade-offs between enhanced safety from poaching and density-dependent effects on reproduction in developing conservation strategies for rhinos and other rare, large species confined to reserve systems.

The higher-taxon approach may provide a pragmatic surrogate for the rapid identification of priority areas for conservation. To date, no continent-wide study has examined the use of higher-taxon data to identify complementarity-based networks of priority areas, nor has the influence of spatial grain size been assessed. We used data obtained from 939 sub-Saharan mammals to analyse the performance of higher-taxon data for continental priority-setting and to assess the influence of spatial grain sizes in terms of the size of selection units (1° × 1°, 2° × 2° and 4° × 4° latitudinal–longitudinal quadrates). Independent of spatial grain size, the selection of priority areas based on genus data is more effective than the random selection of these areas, while the selection of priority areas based on family data is less effective than random selection. Compared to selection based on species data, genus-based priority areas represent between 5–14% fewer species, but represent even fewer of the threatened species (6–31%) and range-restricted species (7–42%) that are normally considered most valuable for conservation. While genus-based priority areas for the 4° grain size represent species almost as effectively as species-based priority areas, genus-based areas perform considerably less effectively than species-based areas for the 1° and 2° grain size. Thus, our results favour the higher-taxon approach for continental priority-setting only when large grain sizes (≥ 4°) are used.

A population and habitat viability analysis (PHVA) was carried out for the Malagasy giant jumping rat (Hypogeomys antimena), one of the key species of the highly threatened dry deciduous forests on the western coast of Madagascar. The occurrence of H. antimena is now confined to 200 km2 of fragmented but relatively undisturbed forest remnants, divided in northern and southern subpopulations. Since 1985 the habitat has declined by about 52% (1985: 42,000 ha; 2000: 20,000 ha). The annual rate of habitat decline increased from 3.9% between 1985 and 1995 to 4.4% between 1995 and 2000. Within a study area (Kirindy Forest/CFPF) in the southern subpopulation, the Hypogeomys population collapsed during the decade 1990-2000, as well as over its remaining range. An ongoing population decline within the next 100 years is predicted even if further habitat decline can be stopped within the next 5 years and no additional mortality by roaming dogs is considered. Thereby, the outcomes of two different individual-based simulation approaches, one that was explicitly written to reflect Hypogeomys life history and VORTEX, one of the most common software packages used for this purpose, were compared. The results of the Hypogeomys model were more pessimistic, but were within the range of predictions of the VORTEX model which calculated larger standard deviations of the mean values. In both subpopulations, a slight increase in the mortality caused by roaming dogs resulted in a predicted decline to an average population size of maximal 170 individuals within 100 years with a high risk of extinction. However, the simulations indicated that the most important factor is reduction in habitat. If the actual annual rate of habitat lost continues, the extinction of the both subpopulations and therefore the entire species is predicted within 24 years.

The natural morphological variation in the wildcat, Felis silvestris, and morphological changes possibly caused by introgressive hybridisation with the domestic cat, F. catus, were examined, based on up to 39 variables concerning cranial morphology. The samples of wild-living cats originated from Scotland and southern Africa and consisted of both classical wildcat and other pelage types. Principal component and cluster analyses suggested that introgressive hybridisation occurred in both areas, with the consequence that the characteristics of local wildcat populations had been altered in terms of the frequencies of occurrence of certain characters, especially those concerning cranial capacity. In both regions the clustering patterns of wild-living cats can be interpreted as containing four main groups. One of these consisted mainly of ‘non-wildcats’ and groups furthest from the ‘non-wild’ cluster consisted of the highest proportion of ‘wildcats’ (c. 80%). We propose that where a population is heavily introgressed, the only feasible way to define a wildcat is on the basis of inter-correlated features and conservationists must take a population-based approach to assess the extent of introgression. This approach may provide an operational standard for assessing the impact of hybridisation between wildcats and domestic cats throughout the species' range; it suggests that the Scottish wildcats may be critically endangered.

Maintenance of genetic diversity is critical for the preservation of small populations. Most captive populations are small and managers focus their efforts on reducing extinction risks by minimising loss of genetic diversity. To project changes in genetic diversity, managers use a genetic drift model. We used data from 40 managed captive populations to test whether this drift model is a good predictor of change in genetic diversity by comparing projected 10-year trajectories of annual loss in expected heterozygosity with the changes estimated from pedigree analysis. We found that the drift model frequently overestimated the rate at which genetic diversity would be lost and could not predict increases, which we observed in 22 species. We suggest that the drift model does not adequately reflect the potential for management strategies, such as prioritised breeding recommendations and recruitment of potential founders, to slow the loss of genetic diversity. In many cases, the drift model may be overly pessimistic, leading managers to seek solutions such as importation, increased population size or the abandonment of a captive population, that may be unnecessary or premature. Managers should be cautious when interpreting results from genetic drift models, taking care to assess how management actions might lead to more optimistic prognoses.

We investigated the distribution of red squirrel (Sciurus vulgaris) populations on the Isle of Wight using Generalized Linear Models (GLMs) and a spatially explicit population dynamics model (SEPM) with the aim of identifying management options for red squirrel conservation. The presence/absence of red squirrels in habitat blocks in the landscape was used as the dependent variable and the size and separation of habitat blocks as independent variables in the GLM analysis. Red squirrel presence within a woodland was significantly related to woodland size and the distance to the nearest wood known to contain squirrels. The SEPM-generated predictions on population viability were closest to the field data when a low dispersal distance of 1 km was used. This suggests that squirrel dispersal on the Isle of Wight is restricted. Our results suggest that some large interconnected woodlands should be maintained if red squirrels are to be conserved for the future.

El Salvador has the smallest protected areas system in Central America. High levels of habitat destruction and disturbance throughout the country make the reserve system especially important for conservation of biodiversity. I used complementarity approaches to assess the relative conservation importance of ten reserves in El Salvador. The principal criterion was presence of nationally threatened and endangered bird species. Additional criteria included residency status (resident species were more important than migratory non-breeding species) and regionally endemic species. Montecristo National Park was the most important area, with 42% of all nationally threatened birds present. El Imposible National Park, El Salvador's largest park (5000 ha), scored second, followed by Laguna El Jocotal wildlife refuge and Barra de Santiago wildlife refuge. In all, these four areas provide refuge for 83% of El Salvador's 268 nationally threatened bird species in less than 7600 ha of natural habitat. Because of small reserve sizes, all of El Salvador's threatened birds may need additional protection.

A common ‘rule of thumb' in conservation biology is that populations of more than 100 individuals are not threatened by demographic stochasticity alone. However, the effect of demographic stochasticity on population survival may be strengthened if the population has a non-homogeneous demography. A recently developed measure, the demographic effective population size ND, scales population size by taking the differential contribution of individuals into account. We exemplify the estimation of ND with the current population of the Saimaa ringed seal, comprising about 200 individuals. The estimate of ND is shown to be dependent on how independently individuals can be assumed to reproduce. Completely independent reproduction leads to ND ≈ 250, suggesting a safe future for the population. However, when acknowledging the potential of breeding failure due to habitat limitation and difficulties of dispersal among the different subareas of Lake Saimaa, ND lies in an alarming range from six to 30 individuals. The effective means to increase the ND value are limited. Although subdivision itself decreases the value of ND, improving dispersal of individuals is shown to be of limited aid. Instead, the most effective way to enhance future prospects of the Saimaa seal population is to improve pup survival to maturity.