A long term study of small mammal populations in a Brazilian agricultural landscape

Long-term monitoring of small mammal populations is very important to understand the variations in temporal abundance on a large time scale, which are related to ecological, economic and epidemiological phenomena. The aim of this study is to monitor the populations of the marsupials Didelphis aurita and Philander frenatus and the rodents Nectomys squamipes, Akodon cursor and Oligorysomys nigripes in a locality of typical Brazilian rural landscape, Sumidouro Municipality, Rio de Janeiro State, Brazil. A mark-recapture study was conducted during five years. We analyzed the population dynamics, the reproduction and age structure of these species. Both marsupials presented higher population sizes in the end of wet period and beginning of the dry period, which can be explained by the seasonal reproduction which begins in the middle of the dry period and ends in the last months of the wet period. N. squamipes reproduced throughout the year but mostly during rainy periods, due to the close association of this rodent to resources found in the water. Higher survivorship and recruitment rates were in the end of the wet season. The rodent A. cursor had an opportunist reproduction, resulting in high turnover rates. Survivorship increased with the effects of the dry periods. O. nigripes showed a clear annual pattern of population cycle with peaks during the dry season. The rodents did not show potential to present outbreaks and become agricultural pests. The annual population cycles of O. nigripes and the unique peak of A. cursor population during five years highlight attention to their importance as wild reservoirs of the hantavirus disease. Their ecological characteristics associated to their opportunistic behavior make these species prone to be good reservoirs of zoonoses.     

Abundance,population structure and growth of the Atrina maura (Bivalvia: Pinnidae) in a tropical lagoon of the Mexican Pacific coast

The abundance, population structure, and growth of the Pen-shell Atrina maura in the Corralero-Alotengo tropical lagoon system in Oaxaca, Mexico, were studied from February to September of 1997. An abundance analysis showed significant temporal and spatial differences (Kruskal-Wallis, p < 0.001). Two spatial groups of abundance were found in the area, one from the mouth of the lagoon system to the middle of the Pen-shell bank, and the other between the middle of the Pen-shell bank and the head of the system. Three temporal periods of abundance were found (February-April-May; July-August; March-June-September). The distribution of population size showed that benthic recruitment of A. maura occurred from February to July. Length frequency of A. maura is commonly wide-ranging; nevertheless, in this study Pen-shell organisms with a valve length of 15 cm were frequently found. The growth rate length was 3.7 cm/month during the dry season (February to May), and 3.5 cm/month during the rainy season (June to September). Sex ratio was maintained at 1:1 from February to May, but males were dominant from June to August, and the minimum length for reproduction was registered at 10 cm valve length.     

Population growth of a gorgonian coral: equilibrium and non-equilibrium sensitivity to changes in life history variables

Summary A size dependent model of population growth of the Caribbean gorgonian Plexaura A is developed based on observed rates of survival, growth and colony fragmentation at a site in the San Blas Islands, Panama. Sensitivity and elasticity analyses indicate that the fate of large colonies has the greatest effect on population growth. Variables which directly affect the generation of large colonies have the next greatest effect on population growth. These variables include the recruitment of large fragments, and the survivorship of colonies in the next smaller size class. Sexual reproduction has an extremely limited ability to affect population growth. Vegetative reproduction has a greater potential effect on growth rates. Environmental conditions regularly change the matrix of transition probabilities which predicts population growth. This keeps the population from approaching its stable size class distribution. Deviations from the stable size class distribution alter sensitivity and elasticity and in this case have the effect of increasing the importance of survivorship of the smallest colonies. Nonequilibrium conditions alter sensitivity analyses and it is important to assess whether populations are at equilibrium and to determine the effects of such deviations on the sensitivity analysis.     

Reproduction and growth in three species of West African fruit bats

This study outlines the reproduction periods of the African fruit bats Epomops buettikoferi, Epomophorus garnbianus , and Micropteropus pusillus at two West African savanna sites: a wet southern Guinea savanna and a drier southern Sudanese savanna. At both sites the two annual birth periods were timed such that both lactation by only one weaning period coincided with the rainy season peaks in fruit availability. On this basis we propose that lactation rather than weaning was the important determinant of the timing of reproduction. There was no evidence of cycling of testes size in males corresponding with the seasonal mating periods. E. buettikoferi and M. pusillus females mated at six months and gave birth at the age of 12 months. Males of the two species reached puberty by 11 months and seven months, respectively. Growth rates did not differ between cohorts growing through the wet and dry season and E. buettikoferi and M. pusillus grew at rates of 276 mg/day and 116 mg/day, respectively.     

Small mammal populations of an agroecosystem in the Atlantic Forest domain, southeastern Brazil.

This study reports 2 years of the population dynamics and reproduction of a small mammal community using the removal method. The study was conducted in a rural area of the Atlantic Forest, in Sumidouro, Rio de Janeiro State, Brazil. The population sizes, age structure and reproduction were studied for the four most common species in the study area. The overall diversity was 1.67 and ranged between 0.8 to 1.67. The species richness was 13 considering the whole study. The most abundant species were the rodents Nectomys squamipes (n = 133), Akodon cursor (n = 74), Oligoryzomys nigripes (n = 25) and the marsupials Didelphis aurita (n = 58) and Philander frenatus (n = 50). Seven other rodents were captured once: Necromys lasiurus, Akodon montensis, Sooretamys angouya, Oecomys catherine, Oxymycterus judex, Euryzygomatomys spinosus and Trinomys iheringi. There were higher peaks for diversity and species richness during the winter (dry) months, probably due to higher food availability. The marsupials had a seasonal reproduction with highest population sizes at the end of the rainy seasons. Nectomys squamipes reproduced mostly during rainy periods. Akodon cursor reproduced predominantly in the winter with the highest population peaks occurring during this season. The analysis of the population dynamics of the rodent species indicated that no species behaved as an agricultural pest, probably due to the heterogeneous landscape of high rotativity of vegetable cultivation. Rodent populations were more susceptible to the removal procedure than marsupial ones.     

Growth patterns of a stream vertebrate differ between urban and forested catchments

1. Urban development results in the decline of amphibian density and species richness. A logical next step towards understanding why urbanisation negatively impacts amphibians is to track species‐specific demographic responses to urbanisation. 2. We monitored growth of two‐lined salamander (Eurycea cirrigera Green) larvae over two complete cohorts (2006 and 2007) in nine western Georgia, U.S.A. streams. 3. We found that salamanders in streams surrounded by urbanised and developing catchments hatched at the same size as their reference‐stream counterparts, but achieved larger sizes within the first few months of growth. We evaluated the effect of four variables that correlate with the urban‐forest gradient and found that elevated temperatures in the urban environment, coupled with decreased intraspecific competition because of lower survivorship in these same habitats, were two of the most likely explanations for increased growth rates. 4. Such an increase in growth of surviving larvae may maintain population viability in urban areas where it has been shown survival is difficult because of increased in‐channel flow during flood events. Because larvae that do survive in urban streams undergo metamorphosis at large sizes, they may recoup a component of fitness (i.e. increased adult survivorship and reproduction) through growth.     

Anopheles subpictus Grassi: Observations on survivorship and population size using mark-release-recapture and dissection methods

A mark-release-recapture experiment to estimate population survivorship and absolute size was performed with wild-caught An. subpictus adults at the village of Khano-Harni, Lahore District, Punjab Province, Pakistan during September 1978, the end of the monsoon rainy season, when temporal population abundance was maximized. Daily survival rate estimated from the recapture sequence of marked adults was low, males=0.192 and females=0.343. Survivorship for females estimated by several vertical age-grading procedures ranged from 0.347 to 0.628. Both stage- and age-specific life tables were calculated from vertical age-grading data determined by the dilatation method. Female and male population size was estimated byBailey 's modification of theLincoln Index and was found to average 4478.4 and 6106.8, respectively. The bionomics, survivorship and population size of An. subpictus in the Lahore are indicated that this species was probably not important in the transmission of human malaria.     

Effects of harvesting onAscophyllum nodosum (L.) Le Jol. (Fucales,Phaeophyta): a demographic approach

Although populations ofAscophyllum nodosum are harvested commercially, little is known about the effects on demographic vital rates (growth, reproduction, survival). This study examines the effects of harvesting season and harvesting intensity on growth, reproduction and mortality of intact fronds in four size classes and in fronds truncated by the harvest. Knowledge of size-specific vital rates was used to evaluate the response of the population to harvesting.Harvesting season and harvesting intensity did not exert a significant effect on growth. Growth in plots not subject to harvesting was less than in harvested plots. No major differences in growth, reproduction and survival between intact and severed fronds emerged. The number of fronds attaining reproduction was enhanced by increased harvesting intensity and by cutting in summer. Harvesting did not seem to induce breakage, and breakage appeared higher in the uncut plots. Most harvesting treatments did not influence survivorship and survivorship was similar among all size classes. Growth rates were inversely related to sizes of fronds.Assessment of variation across size classes yielded more accurate estimates of growth rates than those of previously used methods. Accurate size class specific-growth rates will be a useful criterion when regulating intervals between harvests. Furthermore, assessment of size-specific vital rates allows identification of the frond size classes most relevant to the preservation of resources. Because of their fast growth rates and abundance, fronds in class 1, and, to a lesser extent, class 2, are responsible for most of the population regrowth after harvest. In contrast, classes 3 and 4 contribute little to recovery. This finding provides a strong basis for a harvesting strategy that targets the largest fronds.Author for correspondence     

Effects of ENSO and Temporal Rainfall Variation on the Dynamics of Successional Communities in Old-Field Succession of a Tropical Dry Forest

The effects of temporal variation of rainfall on secondary succession of tropical dry ecosystems are poorly understood. We studied effects of inter-seasonal and inter-year rainfall variation on the dynamics of regenerative successional communities of a tropical dry forest in Mexico. We emphasized the effects caused by the severe El Niño Southern Oscillation (ENSO) occurred in 2005. We established permanent plots in sites representing a chronosequence of Pasture (abandoned pastures, 0–1 years fallow age), Early (3–5), Intermediate (8–12), and Old-Growth Forest categories (n = 3 per category). In total, 8210 shrubs and trees 10 to 100-cm height were identified, measured, and monitored over four years. Rates of plant recruitment, growth and mortality, and gain and loss of species were quantified per season (dry vs. rainy), year, and successional category, considering whole communities and separating seedlings from sprouts and shrubs from trees. Community rates changed with rainfall variation without almost any effect of successional stage. Mortality and species loss rates peaked during the ENSO year and the following year; however, after two rainy years mortality peaked in the rainy season. Such changes could result from the severe drought in the ENSO year, and of the outbreak of biotic agents during the following rainy years. Growth, recruitment and species gain rates were higher in the rainy season but they were significantly reduced after the ENSO year. Seedlings exhibited higher recruitment and mortality rate than sprouts, and shrubs showed higher recruitment than trees. ENSO strongly impacted both the dynamics and trajectory of succession, creating transient fluctuations in the abundance and species richness of the communities. Overall, there was a net decline in plant and species density in most successional stages along the years. Therefore, strong drought events have critical consequences for regeneration dynamics, delaying the successional process and modifying the resilience of these systems.     

Benefits of adjacent habitat patches to the distribution of a crayfish population in a hydro-dynamic wetland landscape

Aquatic macrophyte patches are natural features of wetland ecosystems that serve as habitat for aquatic animals. Previous studies suggest animal densities in Everglades, USA, wetlands are generally less numerous in sawgrass ridges than in deeper lily sloughs. We studied the density distribution of a population of Procambarus fallax in ridge and slough habitat types over a 2-year period, spanning two wet–dry cycles and estimated growth and survival rates under flooded conditions to understand comparative value of each to the crayfish population. Procambarus fallax individuals inhabited and recruited in both marsh habitats. During periods of high water, crayfish densities were similar in both habitats; however, densities in both habitats varied seasonally, leading us to postulate some degree of population redistribution in response to fluctuating water depths. Analysis of size distributions over time revealed juveniles in both habitats and two major recruitment periods each year; distinct juvenile cohorts were present in early winter (Nov–Dec) and mid-summer (July–Aug). An in situ experiment of juvenile growth demonstrated that slough habitat type supported faster growth over ridge habitat. To understand habitat-specific mortality risk, a tethering study during flooded conditions indicated that relative predation risk by aquatic predators was greater in sloughs for all sizes and higher for smaller individuals in both habitats. The comparative importance of ridge and slough balances growth potential and survival probability during flooded conditions. This is the first study through time and across both habitat types analyzing the distribution and size structure of P. fallax population in the Everglades.     

Effects of reproductive mode on demography and life history in Arabis fecunda (Brassicaceae)

Life history theory predicts that trade-offs between growth and reproduction should be dictated by a population's mortality schedule. We tested this prediction with Arabis fecunda, a short-lived perennial that occurs in many different habitats in southwest Montana. Individuals produce either or both axillary or terminal inflorescences. Axillary-flowering plants are usually iteroparous and have smaller reproductive bouts, while terminal-flowering plants have larger reproductive bouts, and tend to be semelparous. We recorded size and fecundity of A. fecunda individuals from 1989 to 1993 in three different habitats. There was great variation in demographic and life history traits among the populations. A wide range of life history strategies among populations of A. fecunda is achieved through different proportions of axillary- and terminal-flowering plants. Arabis fecunda demonstrated a lower recruitment rate, higher survivorship, slower growth, and lower annual fecundity at the low-elevation site compared to the high-elevation site. At the low-elevation site population size was more stable, and elasticity analysis of matrix projection models indicated that adult survivorship was the most important demographic parameter contributing to population growth. This association of life history characters conforms to theoretical predictions.     

Effect of reproductive modes and environmental heterogeneity in the population dynamics of a geographically widespread clonal desert cactus

The dynamics of plant populations in arid environments are largely affected by the unpredictable environmental conditions and are fine-tuned by biotic factors, such as modes of recruitment. A single species must cope with both spatial and temporal heterogeneity that trigger pulses of sexual and clonal establishment throughout its distributional range. We studied two populations of the clonal, purple prickly pear cactus, Opuntia macrocentra, in order to contrast the factors responsible for the population dynamics of a common, widely distributed species. The study sites were located in protected areas that correspond to extreme latitudinal locations for this species within the Chihuahuan Desert. We studied both populations for four consecutive years and determined the demographic consequences of environmental variability and the mode of reproduction using matrix population models, life table response experiments (LTREs), and loop and perturbation analyses. Although both populations seemed fairly stable (population growth rate, λ∼1), different demographic parameters and different life cycle routes were responsible for this stability in each population. In the southernmost population (MBR) LTRE and loop and elasticity analyses showed that stasis is the demographic process with the highest contributions to λ, followed by sexual reproduction, and clonal propagation contributed the least. The northern population (CR) had both higher elasticities and larger contributions of stasis, followed by clonal propagation and sexual recruitment. Loop analysis also showed that individuals in CR have more paths to complete a life cycle than those in MBR. As a consequence, each population differed in life history traits (e.g., size class structure, size at sexual maturity, and reproductive value). Numerical perturbation analyses showed a small effect of the seed bank on the λ of both populations, while the transition from seeds to seedlings had an important effect mainly in the northern population. Clonal propagation (higher survival and higher contributions to vital rates) seems to be more important for maintaining populations over long time periods than sexual reproduction.     

Anopheles culicifacies giles: Adult ecological parameters measured in rural punjab province,pakistan using capture-mark-release-recapture and dissection methods,with comparative observations onAn. stephensi liston andAn. subpictus grassi

Dispersal, immigration and emigration rates, horizontal and vertical survivorship and absolute population size were estimated for micropopulations of An. culicifacies, An. stephensi and An. subpictus at a series of cattle sheds in rural Punjab Province, Pakistan, during November 1979 and May 1980 using capture-mark-release-recapture and dissection methods. Dispersal was temperature-related, with populations more vagile during May. Mean dispersal distance per individual was low for all species. More than 70% of all recaptures were taken at the point of release and the longest detected flight was 1250 meters. Horizontal survivorship was greater during November and was always less than vertical survivorship calculated from dissection agegrading data. Survivorship during the nulliparous period was greater than survivorship throughout total life, indicating the survivorship curve may be slightly sigmoid. Daily population sizes of endemic and immigrating females and males were calculated usingBailey 's (1952) modification of the Lincoln Index, with the daily captures adjusted for immigration which was highest in May. Daily additions to the indoor resting population exclusive of immigrants were estimated using the method ofManly andParr (1968). The relationship of the present findings to malaria transmission and genetic control were discussed.     

Population biology and breeding cycle of the burrowing shrimp Callichirus seilacheri (Decapoda,Callianassidae) from the eastern tropical Pacific

Year-round continuous reproduction in tropical regions is an established paradigm in marine ecology. In this study, we tested this paradigm using the ghost shrimp Callichirus seilacheri from the tropical eastern Pacific as a model species. We also examined size-frequency distribution, sex ratio, and recruitment cycle to contribute to the biological knowledge of this species. To this end, a total of 456 individuals of C. seilacheri were collected during 12 months of sampling. Population structure was symmetrical for both sexes, and the overall sex ratio did not differ from evenness. Males outnumbered females in smaller size classes, though, revealing a potential sex-dependent mortality in small individuals. The breeding pattern followed the well-marked seasonal regime of the region, with ovigerous females registered during the rainy season. While natural variation in the seawater temperature had no influence on reproduction of this species, changes in water salinity possibly triggered the appearance of egg-bearing females in the population. Recruitment occurred throughout the year but was more intense during the dry season, following the appearance of ovigerous females. The adaptability of the life cycle of C. seilacheri to the seasonal climate provides further evidence that reproduction in tropical species is not always continuous.     

Recruitment and survival of native annual Trifolium species in the highlands of Ethiopia

Germination and survival of indigenous annual Trifolium species were studied. Seedlings naturally emerging in the field and from sown seeds in pots were regularly counted and uprooted, and survival was studied by monitoring colour‐coded seedlings. Differences in recruitment of Trifolium species were strongly related to the rainfall pattern. In the fallow (crop) lands, no seedling survived the dry season between the short and main rainy periods during the year, while in the natural pasture, 8% of the seedlings survived into the main growing (rainy) season. Occasional rains occurring in the dry period (between the two rainy periods) also induced successions of germination and seedlings death, and therefore depleted the soft seed reserve in the soil by the beginning of the main growing season. As temperature fluctuations were minimal during the main rainy season, the rate of seed softening was low, affecting new germinations. This had a significant impact on the quality of the natural pastures on which livestock in the highlands are dependent.     

Demography of palm species in Brazil’s Atlantic forest: a comparison of harvested and unharvested species using matrix models

Surprisingly little is known about the effect of alternative harvesting methods and different means of reproduction on the population dynamics of plant species. Here, we test the hypothesis that habitat fragmentation leads to negative population growth rates for three palm species in Brazil’s biodiversity-rich and highly fragmented Atlantic forest. We compared the demography (λ and elasticities) of Astrocaryum aculeatissimum, Euterpe edulis and Geonoma schottiana in five fragments ranging in size from 19 to 3500 ha. A total of 4.05 ha were censused throughout the landscape studied. All individuals of the three palms were tagged in 2005 and their survival was monitored in 2006 and 2007. All new plants were also recorded. Summary matrices were constructed for each transition years by pooling data from all plots of all fragments together for each species. Based on summary matrices, A. aculeatissimum and G. schottiana asymptotic population growth rates were not significantly different from 1, suggesting that populations of these palms were stable. The projection of E. edulis populations was to decrease in size, with 95% confidence intervals of λ for these estimates failing to exceed one in the second transition year. A. aculeatissimum, E. edulis and G. schottiana survivorship was high for all stage classes and exceeded 85% for postseedling stages. We found that the more important vital rates were the survival rates in larger stages for all three species. Our results show that the maintenance of high levels of reproductive survivorship should be a focus of conservation strategies for these species, especially in the case of large E. edulis individuals.     

Life-history models of extinction: a test with island spiders

This study analyzes extinction patterns for two species of orb spiders monitored annually on 77 islands over a continuous 20-yr period. One species, Argiope argentata, has large populations sometimes crashing quickly to extinction and a much weaker relation of extinction likelihood to population size than does the other species, Metepeira datona. Demographic models were built for both species and matched against observations. Differences between the species in life-history traits-estimated with measurements from the field-together with incorporation of demographic stochasticity, a population ceiling, and environmental stochasticity, were necessary to fit the observed extinction curves. As predicted from life-history patterns, long-term population growth rates (and hence predicted extinction probabilities) are relatively very sensitive to values of juvenile survivorship. Models are also sensitive to variation in the population ceiling and environmental noise, which tend to act in a complementary manner. A simple model with no age structure was able to fit the data on large initial population sizes but not on small initial population sizes, showing that life cycle characteristics interact with the various sources of stochasticity and hence have to be taken into account to produce a precise model of the extinction process.     

Responses to Fire in Selected Tropical Dry Forest Trees1

Fire is a frequent disturbance in the tropical dry forests of Central America, yet very little is known about how native species respond to such events. We conducted an experimental burn in a tropical dry forest of western Nicaragua to evaluate plant responses to fire with respect to survivorship and recruitment. Measurements of woody vegetation of all size classes were carried out prior to the prescribed burn and three successive years post fire. We selected the 15 most abundant species <10 cm DBH to assess percent survivorship and sprouting responses post fire. Changes in seedling densities for these 15 most abundant species and the 15 least abundant species were analyzed using a repeated measure ANOVA. We also assessed changes in seedling densities for three species of international conservation concern. We found three major fire‐coping strategies among common dry forests plants: resisters (low fire‐induced mortality), resprouters (vigorous sprouting), and recruiters (increased seeding post‐fire). While survivorship was generally high relative to tropical moist forest species, those species with lower survivorship used either seeding or sprouting as an alternative strategy for persisting in the forest community. Seed dispersal mechanisms, particularly wind dispersal, appear to be an important factor in recruitment success post‐fire. Burn treatment led to a significant increase in the density of seedlings for two species of conservation concern: Guaiacum sanctum and Swietenia humilis. Results of this study suggest that common dry forest species in western Nicaragua are fire tolerant. Further study of individual species and their fire responses is merited.     

Gopher Tortoise Demographic Responses to a Novel Disturbance Regime

The long-term viability of gopher tortoise (Gopherus polyphemus) populations is jeopardized by increased urbanization and habitat degradation owing to fire suppression. Because the species' remaining natural habitats in the southeastern United States exist within a mosaic of anthropogenic land uses, it is important to understand demographic responses to contrasting land uses and habitat management regimes. We examined differences in demographic parameters among fire-suppressed sandhill, restored sandhill, and former sandhill (i.e., ruderal) land use-land cover (LULC) types at Archbold Biological Station in south-central Florida, USA. Using Program MARK, we estimated population size, and sex-specific and LULC-specific survivorship based on 6 years of mark-recapture data. We also analyzed individual growth trajectories and clutch sizes to determine whether growth rates or reproductive output differed among LULC types. Tortoises in an open, ruderal field occurred at a higher density (7.79/ha) than in adjacent restored (1.43/ha) or fire-suppressed (0.40/ha) sandhill. Despite this higher density, both adult survivorship and body size were significantly higher in the ruderal field. Furthermore, the larger female body size in the ruderal field likely contributed to increased annual survivorship and slightly larger average clutch sizes. We did not detect offsetting negative demographic effects; in particular, we did not find significant biological or statistical differences in body condition, asymptotic body size, or growth rate among the 3 LULC types. Our results suggest that anthropogenic, grass-dominated land-cover types may be important components of the habitat mosaic currently available to this at-risk species. © 2019 The Wildlife Society.     

Population dynamics of a stingless bee community in the seasonal dry lowlands of Costa Rica

This study examines the dynamics of a population of stingless bee colonies in the seasonal tropics of Guanacaste, Costa Rica. The community in a forest remnant was compared with that in surrounding deforested areas. During this 4-year study, a total of 192 wild stingless bee colonies were recorded, belonging to 14 species. Population dynamics were highly seasonal. Colony mortality peaked at the end of the wet season (October–November) while colony reproduction was most frequent during the dry season (December to April). Colony survival was not lower in founder colonies compared to established colonies. The most common species, T. angustula, had a much lower probability of annual survivorship in the forest (P = 0.74) than in deforested areas (P = 0.92). This results in an estimated colony life span for T. angustula of 3.8 years in the forest and 12.5 years in deforested areas. T. angustula should swarm once every two years to maintain its forest population, but only once every 12.5 years to maintain its population in the deforested areas. Survivorship of all other stingless bees was similar in the forest and deforested areas and did not significantly differ among the species. The average annual survivorship probability of these species was as high as 0.96, resulting in an estimated colony life span of 23.3 years. On average only one swarm per 20 years is needed to maintain their populations. Life history of the sympatric Africanised honey bee clearly differed from that of the stingless bees, with much lower annual survivorship probabilities for both founder (none survived) and established colonies (P = 0.33). These figures support the general idea that stingless bees invest more in colony survival rather than reproduction, but also show that life history is affected by both species and location. Received 27 October 2004; revised 8 March and 15 June 2005; accepted 5 July 2005.