Search

Many wildlife populations are either naturally, or as a result of human land use, patchily distributed in space. The degree of fragmentation-specifically the remaining patch sizes and habitat configuration-is an important part of population dynamics. Demographic stochasticity is also likely to play an important role in patchy habitats that host small local populations. We develop a simulation model to evaluate the significance of demographic stochasticity and the role fragmentation plays in the determination of population dynamics and the risk of extinction of populations on habitat patches. Our model is formulated as a Markov-chain stochastic process on a finite, spatially explicit array of patches in which probability of successful dispersal is a function of interpatch distance. Unlike past work, we explicitly model local population dynamics and examine how these scale up to the entire population. As a test case, we apply the model to the American pika (Ochotona princeps) population living on the ore dumps in the ghost mining town of Bodie, California. This population has been studied nearly continuously for over four decades and has been of conservation concern as the southern half of the population declined precipitously beginning in 1989. Our model suggests that both the specific configuration of habitat and landscape heterogeneity are necessary and sufficient predictors of the eventual extinction of the southern constellation of patches. This example has important implications, as it suggests that fragmentation alone can lead to regional extinctions within metapopulations.

While there is widespread recognition of human involvement in biodiversity loss globally, at smaller spatial extents, the effects are less clear. One reason is that local effects are obscured by the use of summary biodiversity variables, such as species richness, that provide only limited insight into complex biodiversity change. Here, we use 30 yr of invertebrate data from a metacommunity of 10 streams in Wales, UK, combined with regional surveys, to examine temporal changes in multiple biodiversity measures at local, metacommunity, and regional scales. There was no change in taxonomic or functional a-diversity and spatial b-diversity metrics at any scale over the 30-yr time series, suggesting a relative stasis in the system and no evidence for on-going homogenization. However, temporal changes in mean species composition were evident. Two independent approaches to estimate species niche breadth showed that compositional changes were associated with a systematic decline in mean community specialization. Estimates of species-specific local extinction and immigration probabilities suggested that this decline was linked to lower recolonization rates of specialists, rather than greater local extinction rates. Our results reveal the need for caution in implying stasis from patterns in a-diversity and spatial b-diversity measures that might mask non-random biodiversity changes over time. We also show how different but complementary approaches to estimate niche breadth and functional distinctness of species can reveal long-term trends in community homogenization likely to be important to conservation and ecosystem function.

Where predation is seasonally variable, the potential impact of a predator on individual prey species will critically depend on phenological synchrony of the predator with the prey. Here we explored the effects of seasonally variable predation in multispecies assemblages of short-lived prey. The study was conducted in a landscape in which we had previously demonstrated generally high, but spatially and seasonally variable dragonfly-induced mortality in adult butterflies. In this system, we show that patterns of patch occupancy in butterfly species flying during periods of peak dragonfly abundance are more strongly associated with spatial variation in dragonfly abundance than patch occupancy of species flying when dragonfly density was low. We provide evidence indicating that this differential sensitivity of different butterfly species to between-habitat differences in dragonfly abundance is causally tied to seasonal variation in the intensity of dragonfly predation. The effect of dragonfly predation could also be measured at the level of whole local butterfly assemblages. With dragonfly density increasing, butterfly species richness decreased, and butterfly species composition tended to show a shift toward a greater proportion of species flying during periods of off-peak dragonfly abundance.

Habitat fragmentation often arises from human-induced alterations to the matrix that reduce or eliminate dispersal between habitat patches. Elimination of dispersal increases local extinction and decreases recolonization. These phenomena were observed in the eastern collared lizard (Crotaphytus collaris collaris), which lives in the mid-continental highland region of the Ozarks (Missouri, USA) on glades: habitats of exposed bedrock that form desert-like habitats imbedded in a woodland matrix. With the onset of woodland fire suppression, glade habitats degenerated and the woodland matrix was altered to create a strong barrier to dispersal. By 1980, lizard populations in the Ozarks were rapidly going extinct. In response to this decline, some glades were restored by clearing and burning. Starting in 1984, collared lizard populations were translocated onto these restored habitats. The translocated populations persisted but did not colonize nearby glades or disperse among one another. In 1994 prescribed woodland fires were initiated, which unleashed much dispersal and colonizing behavior. Dispersal was highly nonrandom by both intrinsic variables (age, gender) and extrinsic variables (overall demography, glade population sizes, glade areas, landscape features), resulting in different classes of lizards being dominant in creating demographic cohesiveness among glades, colonizing new glades on a mountain, and colonizing new mountain systems. A dramatic transition was documented from isolated fragments, to a nonequilibrium colonizing metapopulation, and finally to a stable metapopulation. This transition is characterized by the convergence of rates of extinction and recolonization and a major alteration of dispersal probabilities and pattern in going from the nonequilibrium to stable metapopulation states.

There is considerable debate among ecologists as to whether or not communities are saturated. In saturated communities, species richness should remain relatively constant over time, despite compositional turnover, because richness is negatively correlated with colonization and positively correlated with local extinction. Few studies have tested for saturation using temporal observational data as well as diversity-perturbation experiments. We analyzed 10 years of data for plant species richness at 71 sites on contrasting serpentine and non-serpentine soils within Californian (USA) grasslands. We also manipulated local richness and measured its effects on immigration and extinction. Consistent with saturation, we observed that richness was positively correlated with extinction rates and negatively correlated with colonization rates, and randomization tests confirmed that diversity fluctuated less than expected by chance. However, experimental species additions and removals did not affect extinction or colonization, suggesting that richness is not regulated by local species interactions. Instead, we propose three reasons why richness may fluctuate within narrow limits causing the appearance of saturation in temporal observational data sets: negatively autocorrelated patterns of biotic response to yearly conditions, differential affinities of particular species for local conditions, or stochastic abundance-dependent colonization and extinction rates. We illustrate the latter using a metacommunity model.

Dispersal limitation plays an important role in a number of equilibrium and nonequilibrium theories about community ecology. In this study we use the framework of island biogeography to look for evidence of dispersal limitation in ectomycorrhizal fungal assemblages on "tree islands," patches of host trees located in a non-host vegetation matrix. Because of the potentially strong effects of island area on species richness and immigration, we chose to control island size by sampling tree islands consisting of a single host individual. Richness on tree islands was high, with estimates ranging up to 42 species of ectomycorrhizal fungi associating with a single host individual. Species richness decreased significantly with increasing isolation of tree islands, with our regression predicting a 50% decrease in species richness when tree islands are located distances of approximately 1 km from large patches of contiguous forests. Despite the fact that fungal fruit bodies produce large numbers of spores with high potential for long-distance travel, these results suggest that dispersal limitation is significant in ectomycorrhizal assemblages. There were no discernible effects of isolation or environment on the species identity of tree island fungal colonists. In contrast to the highly predictable patterns of tree island colonization we observed in a previous study on early successional forests, we suggest that over longer time periods the community assembly process becomes more dominated by stochastic immigration and local extinction events.

The forest fires induced by the El Niño Southern Oscillation (ENSO) in 1997-1998 resulted in the temporary extirpation of more than 100 lowland butterfly species at a forest site in Borneo. Species with more restricted ranges were less likely to recover over the following four years. Matched-pair analyses revealed that species with lower initial abundances, restricted geographic ranges, and more specialized larvae were less likely to return. Specialization differed predictably between the (more generalist) wide-range and (more specialized) restricted-range species in our data set, and both geographic range and level of specialization were important in multivariate models. These are the first observations directly linking extent of occurrence, ecological specialization, and observed recovery following local extirpation. If recovery time exceeds the frequency of disturbance, local extirpation can lead to local extinction. Given that ENSO-induced disturbances are increasing in frequency, in severity, and in geographic scale, these results suggest that specialist species with restricted geographic ranges could be at particularly high risk of global extinction.

Isolation of remnant forest patches, coupled with anthropogenic changes in the surrounding landscape, often leads to changes in the population density of forest-dwelling mammals. At Pasoh Forest Reserve in Peninsular Malaysia, densities of native wild pigs (Sus scrofa) are 10–100 times greater than historical levels due to the local extinction of feline predators and the presence of abundant food sources in areas adjacent to the forest. Female pigs build reproductive nests out of understory vegetation; at Pasoh these nests are constructed primarily of woody saplings several meters tall that pigs snap or uproot, causing substantial local damage. We documented the prevalence of nest building by pigs at Pasoh within a 25-ha area from 1995 to 1998 and investigated the impacts of this mammalian behavior on the understory plant community. In total, 643 pig nests were enumerated, providing an estimate that 6.0 nests·ha−1·yr−1 were constructed in the survey area. Pigs avoided constructing nests adjacent to trails but otherwise built nests throughout the 25-ha survey area. A single pig nest contained, on average, 267 ± 86 (mean ± 1 sd) woody saplings, of which 45% had been uprooted and 55% had the main stem snapped. The understory area affected by the construction of a single pig nest averaged 244 ± 112 m2, with 53% of all free-standing woody plants ≥70 cm tall and

Metapopulation models assume that local population size (in a habitat patch) and therefore local extinction probability, is a function of patch size, and that interpatch movement rate and therefore recolonization of local extinctions is a function of both patch size and patch isolation. We hypothesized that the predictive power of models that relate patch immigration rate to patch size and isolation will be reduced when spatial structure in the landscape matrix (the nonhabitat portion of the landscape) affects organism movement through the landscape. We used a simulation model to evaluate this hypothesis for three different aspects of matrix spatial structure (contrast among matrix cover types, number of matrix cover types, and grain of matrix spatial pattern) and for two different types of movement behavior, representing a specialist and a generalist species. We tested the hy- pothesis for one aspect of matrix structure (number of matrix cover types) in a translocation field study of the eastern chipmunk ( Tamias striatus, a habitat specialist) and the white- footed mouse (Peromyscus leucopus, a habitat generalist). When the matrix was composed of a small number of cover types, patch size and isolation accounted for up to 75% of the variation in patch immigration rate in the simulation study, and for 61% of the variation in interpatch movement in the field study. However, when the matrix was composed of a large number of cover types, the amount of explained variation dropped to as little as 33% for the simulation study and to 17% in the field study. Our results suggest that patch characteristics, such as patch size and isolation, may be poor predictors of interpatch movement when the landscape matrix is heterogeneous and when the organism responds to boundaries between different matrix cover types. These results imply that habitat patch- based models, such as those based on current metapopulation theory, will perform poorly in these situations.

Few species are likely to be so evident that they will always be detected when present. Failing to allow for the possibility that a target species was present, but undetected, at a site will lead to biased estimates of site occupancy, colonization, and local extinction probabilities. These population vital rates are often of interest in long-term monitoring programs and metapopulation studies. We present a model that enables direct estimation of these parameters when the probability of detecting the species is less than 1. The model does not require any assumptions of process stationarity, as do some previous methods, but does require detection/nondetection data to be collected in a manner similar to Pollock's robust design as used in mark-recapture studies. Via simulation, we show that the model provides good estimates of parameters for most scenarios considered. We illustrate the method with data from monitoring programs of Northern Spotted Owls ( Strix occiden- talis caurina) in northern California and tiger salamanders (Ambystoma tigrinum) in Min- nesota, USA.

We take advantage of a fortuitous local extinction and recolonization of white-lipped peccaries (WLPs) at the Cocha Cashu Biological Station in southeastern Peru to assess the impact of this high-biomass seed predator on the recruitment of a dominant member of the tree community, the palm Astrocaryum murumuru. WLPs were common at Cocha Cashu in the mid-1970s. In 1978, the species vanished from the entire region and did not reappear until 1990. To assess the impacts of the presence-absence of WLPs on Astrocaryum recruitment, we conducted transect counts of both the number and spatial distribution of palm seedlings in 1978 (when WLPs were present), in 1990 (after a 12-yr absence of WLPs), and in 1999 (after 10 yr of recovery). Other factors affecting recruitment, such as climate variability, differences in tree fecundity, and fluctuations of vertebrate and invertebrate seed predators were also examined as alternative hypotheses for any changes in seedling spatial distribution and abundance. In the absence of WLPs, the density of Astrocaryum seedlings increased 1.7-fold, and the spatial distribution of seedlings with respect to safe sites was significantly altered. After the return of WLPs to the study area, seedling density dropped to its former level, and the spatial distribution of seedlings with respect to safe sites returned to the same pattern found 21 yr earlier in 1978. None of the other factors investigated varied systematically with the census intervals. These results demonstrate that the absence of a single, albeit important, member of a diverse seed predator guild can have a major impact on the demography of a common tree species. We infer that individual seed predators can play central roles in regulating the demography of tropical trees, as envisioned many years ago by Janzen and Connell, and that their extirpation or extinction can have cascading effects in tropical ecosystems.

Theoretical efforts and small-scale experiments have given rise to the widespread belief that the fewer occurrences a species has or the more fragmented its distribution is, the more vulnerable that species should be to extinction. Lacking, however, are large-scale multi-species studies exploring the connection between rarity and local extinction risk. Here we present a landscape-level biogeographic test of this widely assumed linkage. Using a unique data set detailing the occurrence patterns of freshwater fishes of the Sonoran Desert (a gravely endangered fauna) we obtained for each of 25 species a measure of rarity that was independent of spatial scale. We found that fragmentation was consistently associated with elevated extinction risk, whereas the number of occurrences exerted a significant effect only if fragmentation had not already been accounted for. Specifically, desert fish species with the most fragmented historic distributions were nearly five times more likely to suffer local extirpations (since 1980) than were species with more continuous distributions. These findings underscore what a strong link exists between spatial distribution and vulnerability to extinction, clarifying that the link exists even at the landscape level and across an entire biogeographic fauna.

We incorporate stochastic, density-dependent seasonal recruitment in adult Anopheles mosquito populations in a discrete-event model of Plasmodium falciparum malaria transmission and find the probabilities of parasite extinction higher than with perennial transmission. Seasonal fluctuations in vector populations act to synchronize the dynamics of infection and immunity in host populations, leading to fluctuations in parasite prevalence greater than expected solely on the basis of high- and low-season vector densities. This synchronization also biases frequencies of infection with multiple parasite phenotypes or genotypes.

Stream landscapes are highly variable in space and time and, like terrestrial landscapes, the resources they contain are patchily distributed. Organisms may disperse among patches to fulfill life-history requirements, but biotic and abiotic factors may limit patch or locality occupancy. Thus, the dynamics of immigration and extinction determine, in part, the local structure of assemblages. We sampled fishes and stream habitat at 12 localities for two years (96 samples) to examine the deterministic nature of immigration and extinction processes in stream fish assemblages. Mean immigration rates for assem- blages were highest at large stream localities, where the pool of potential immigrants was largest. Mean extinction rates were highest where variability in the flow regime was high, though local refugia appeared to modify the extinction process at one locality. Significant nested subset patterns in species composition occurred over time for 7 of the 12 localities. The strength of the nesting was associated with mean immigration and extinction rates. Higher extinction rates corresponded to stronger nestedness, whereas higher immigration rates were associated with weaker nestedness. Across all species, both immigration and extinction rates were strongly associated with mean abundance. Species with high local abundances had higher immigration rates and lower extinction rates than did species with low local abundances. There were no significant associations between trophic guild or body size and immigration and extinction rate. This work supports the hypothesis that immigration and extinction rates for assemblages are predictable along environmental gradients, and that species are less prone to local extinction and more prone to colonize areas when they maintain high local abundances. The extinction process in local assemblages can be a highly ordered event leading to strong nested subset patterns, but immigration appears to be more stochastic.

Population decline, local extinction, and recovery are profoundly influenced by variation in demography and life-history traits. In open populations, changes in patterns of recruitment may also have a major influence on the size of local populations, particularly for short-lived organisms. We examine here the demographic processes underlying a slow decline of corals on Jamaican reefs, where coral cover has decreased by fourfold over a 16-yr period. We divided the study into three approximately equal intervals (1977–1982, 1982–1987, and 1987–1993) and constructed size-based transition matrices for each of three abundant species of corals (Montastrea annularis, Agaricia agaricites, and Leptoseris cucullata) that differ substantially in life history: Montastrea is slower-growing, longer-lived, and has lower rates of recruitment than the other two species. Rates of survival, population growth (λ), and recruitment declined substantially over time for all species and the stable size structures became increasingly dominated by small colonies. Elasticity and life table response analysis showed that changes in the persistence of large colonies had the biggest impact on population growth in all species. Simulations indicated that the levels of larval recruitment required to maintain populations at 1977 levels increased sharply over time, even as the actual recruitment rate declined. Recruitment failure was much more important to A. agaricites and L. cucullata than to M. annularis, which could survive long periods with minimal larval input. Recovery of these populations will require an increase in both survival and recruitment. The likelihood of the latter will depend on the scale of larval dispersal, and on the impact of large-scale mortality of adults on stock-recruitment relationships. Differences in connectivity and life histories of corals will determine future patterns of recovery or further decline.

Epidemiological theory suggests that pathogens will not cause host extinctions because agents of disease should fade out when the host population is driven below a threshold density. Nevertheless, infectious diseases have threatened species with extinction on local scales by maintaining high incidence and the ability to spread efficiently even as host populations decline. Intertidal black abalone (Haliotis cracherodii ), but not other abalone species, went extinct locally throughout much of southern California following the emergence of a Rickettsiales-like pathogen in the mid-1980s. The rickettsial disease, a condition known as withering syndrome (WS), and associated mortality occur at elevated water temperatures. We measured abalone body temperatures in the field and experimentally manipulated intertidal environmental conditions in the laboratory, testing the influence of mean temperature and daily temperature variability on key epizootiological processes of WS. Daily temperature variability increased the susceptibility of black abalone to infection, but disease expression occurred only at warm water temperatures and was independent of temperature variability. These results imply that high thermal variation of the marine intertidal zone allows the pathogen to readily infect black abalone, but infected individuals remain asymptomatic until water temperatures periodically exceed thresholds modulating WS. Mass mortalities can therefore occur before pathogen transmission is limited by density-dependent factors.

The bush cricket, Metrioptera bicolor Philippi (Orthoptera, Tettigoniidae) lives in a restricted area with grassland patches separated by mainly pine forest. The regional population dynamics can be described as a metapopulation with local extinctions and col- onizations. I used inventory data in order to test the following hypotheses: (1) temporal variability of local population size should be negatively correlated with habitat heteroge- neity, when habitat quality changes in relation to fluctuating weather conditions; (2) ex- tinction risk should be higher on homogeneous habitat patches than on heterogeneous patches. Both hypotheses are convincingly corroborated and I find a compensatory effect between patch area and heterogeneity. Local extinctions may occur on large, but homo- geneous patches, and local populations can survive quite well on small patches that contain a mosaic of several types of grassland vegetation with different humid conditions. I discuss the implications of these findings for the preservation of endangered species that are sus- ceptible to regional stochasticity.

Despite the importance of habitat in determining species distribution and persistence, habitat dynamics are rarely modeled in studies of metapopulations. We used an integrated habitat-occupancy model to simultaneously quantify habitat change, site fidelity, and local colonization and extinction rates for larvae of a suite of Great Plains stream fishes in the Arikaree River, eastern Colorado, USA, across three years. Sites were located along a gradient of flow intermittency and groundwater connectivity. Hydrology varied across years: the first and third being relatively wet and the second dry. Despite hydrologic variation, our results indicated that site suitability was random from one year to the next. Occupancy probabilities were also independent of previous habitat and occupancy state for most species, indicating little site fidelity. Climate and groundwater connectivity were important drivers of local extinction and colonization, but the importance of groundwater differed between periods. Across species, site extinction probabilities were highest during the transition from wet to dry conditions (range: 0.52-0.98), and the effect of groundwater was apparent with higher extinction probabilities for sites not fed by groundwater. Colonization probabilities during this period were relatively low for both previously dry sites (range: 0.02-0.38) and previously wet sites (range: 0.02-0.43). In contrast, no sites dried or remained dry during the transition from dry to wet conditions, yielding lower but still substantial extinction probabilities (range: 0.16-0.63) and higher colonization probabilities (range: 0.06-0.86), with little difference among sites with and without groundwater. This approach of jointly modeling both habitat change and species occupancy will likely be useful to incorporate effects of dynamic habitat on metapopulation processes and to better inform appropriate conservation actions.

Dispersal is known to play a crucial role in the formation of species' ranges. Recent studies demonstrate that dispersiveness increases rapidly during the range expansion of species due to a fitness increase for dispersers at the expanding front. R. D. Holt concluded, however, that emigration should decline after the period of invasion and hence predicted some range contraction following the initial expansion phase. In this study, we evaluate this hypothesis using a spatially explicit individual-based model of populations distributed along environmental gradients. In our experiments we allow the species to spread along a gradient of declining conditions. Results show that range contraction did emerge in a gradient of dispersal mortality, caused by the rapid increase in emigration probability during invasion and selection disfavoring dispersal, once a stable range is formed. However, gradients in growth rate, local extinction rate, and patch capacity did not lead to a noticeable contraction of the range. We conclude, that the phenomenon of range contraction may emerge, but only under conditions that select for a reduction in dispersal at the range edge in comparison to the core region once the expansion period is over.

Perturbation analysis is a powerful tool to study population and community dynamics. This article describes expressions for sensitivity metrics reflecting changes in equilibrium occupancy resulting from small changes in the vital rates of patch occupancy dynamics (i.e., probabilities of local patch colonization and extinction). We illustrate our approach with a case study of occupancy dynamics of Golden Eagle (Aquila chrysaetos) nesting territories. Examination of the hypothesis of system equilibrium suggests that the system satisfies equilibrium conditions. Estimates of vital rates obtained using patch occupancy models are used to estimate equilibrium patch occupancy of eagles. We then compute estimates of sensitivity metrics and discuss their implications for eagle population ecology and management. Finally, we discuss the intuition underlying our sensitivity metrics and then provide examples of ecological questions that can be addressed using perturbation analyses. For instance, the sensitivity metrics lead to predictions about the relative importance of local colonization and local extinction probabilities in influencing equilibrium occupancy for rare and common species.

The interaction between species, mediated by a shared natural enemy (i.e., apparent competition), has been the subject of much theoretical and empirical investigation. However, we lack field experiments that assess the importance of apparent competition to metacommunity structure. Here, I conducted a series of field experiments to test whether apparent competition, mediated by shared egg parasitoids (Anagrus nigriventris and A. columbi), occurs between two abundant planthopper species (Delphacodes scolochloa and Prokelisia crocea) of the North American Great Plains. The two planthoppers feed on different plant species within prairie potholes (wet depressions) and, thus, do not interact directly. At the scale of individual potholes, a five-fold pulse increase in D. scolochloa density (relative to control potholes) resulted in a steady decline in P. crocea density over two generations. As expected in cases of apparent competition, P. crocea eggs in these potholes suffered twice the level of parasitism as P. crocea eggs in control potholes. In contrast, a sixfold increase in P. crocea density had no effect on D. scolochloa density or parasitism in those potholes. The superiority of D. scolochloa over P. crocea likely can be attributed to a larger source population size, greater amount of host habitat, and/or the presence of a phenological refuge from parasitism for D. scolochloa. In another experiment, in which small populations of P. crocea were established either in close proximity to D. scolochloa or in isolation, I found that the likelihood of P. crocea persistence was 36% lower in the former than the latter populations. This difference was attributable to very high rates of parasitism of P. crocea when adjacent to D. scolochloa. These two experiments provide clear evidence that the two planthopper species engage in apparent competition and that the shared parasitoids may play a significant role in limiting membership in a local community. Based on these findings, I argue that metacommunity studies must be broadened to include higher trophic levels.

The global extinction of species proceeds through the erosion of local populations. Using a 60-year time series of annual sighting records of plant species, we studied the correlates of local extinction risk associated with a risk of species extinction in the Park Grass Experiment where plants received long-term exposure to nutrient enrichment, soil acidification, and reductions in habitat size. We used multivariate linear models to assess how extrinsic threats and life history traits influence extinction risk. We investigated effects of four extrinsic threats (nitrogen enrichment, productivity, acidification, and plot size) as well as 11 life history traits (month of earliest flowering, flowering duration, stress tolerance, ruderalness [plant species’ ability to cope with habitat disturbance], plant height, diaspore mass, seed bank, life form, dispersal mode, apomixis [the ability for a species to reproduce asexually through seeds], and mating system). Extinction risk was not influenced by plant family. All of the 11 life history traits except life form and all threat variables influenced extinction risk but always via interactions which typically involved one threat variable and one life history trait. We detected comparatively few significant interactions between life history traits, and the interacting traits compensated for each other. These results suggest that simple predictions about extinction risk based on species’ traits alone will often fail. In contrast, understanding the interactions between extrinsic threats and life history traits will allow us to make more accurate predictions of extinctions.

Past studies of local extinctions in fragmented habitats most often tested the influence of fragment size and isolation while ignoring how differences in the surrounding landscape matrix may govern extinction. We assessed how both the spatial attributes of remnant patches (area and isolation) and landscape factors (extent of urbanization and maximum inter-fire interval) influence the persistence of native plant species in grasslands in western Victoria, Australia. Persistence was determined in 2001 by resurveying 30 remnants first surveyed in the 1980s, and correlates of extinction were assessed using Bayesian logistic regression models. On average, 26% of populations of native species became locally extinct over two decades. Area and isolation had little effect on the probability of local extinction, but urbanization and longer maximum inter-fire intervals increased extinction risk. These findings suggest that the native grasslands studied are relatively insensitive to area- and isolation-based fragmentation effects and that short-term persistence of plant populations requires the maintenance of habitat quality. The latter is strongly influenced by the landscape matrix surrounding remnant patches through changes in fire regimes and increased exogenous disturbance.

We integrated across individual, population, community, and ecosystem levels to understand the impact of environmental stress by tracking the foundation species Mytilus edulis in the hypoxic estuary Narragansett Bay, Rhode Island, USA. Our initial surveys revealed that the mussels occurred in nine extensive (2-28 ha) dense (814-9943 individuals/m2) subtidal reefs that attracted a diverse suite of predators (sea stars, crabs, gastropods). Hypoxia occurred in the summer of 2001, and a mussel transplant experiment revealed overall reduced growth rates of individuals, and higher mortality rates among larger mussels. At the population level, large decreases in densities and cover of mussels were correlated with dissolved oxygen concentrations, leading to extinction at one site and reductions of over an order of magnitude at others. Within one year, seven of the eight remaining populations were edged to extinction, and the previously extinct population was recolonized. At the community level, a predator exclusion experiment indicated that predation was an unimportant source of mussel mortality during the hypoxic period, in part due to the emigration of sea stars, as predicted by the Consumer Stress Model. However, mussels were too intolerant to hypoxia to have a net benefit from the predation refuge. The seasonal (summer) occurrence of hypoxia allowed sea stars to return following a lag, as predicted by a stress return time model, and the resumption of predation contributed to the subsequent extinction of mussel populations. At the ecosystem level, the initial filtration rate of the mussel reefs was estimated at 134.6 x 10(6) m3/d, equivalent to filtering the volume of the bay 1.3 times during the 26-d average residence time. That function was reduced by >75% following hypoxia. The effect of hypoxia on each level of organization had consequences at others. For example, size-specific mortality and decreased growth of individuals, and reduced filtration capacity of reefs, indicated a loss of the ability of mussels to entrain planktonic productivity and potential to control future eutrophication and hypoxia. Our study quantified patterns of loss and identified pathways within an integrative framework of feedbacks, summarized in a conceptual model that is applicable to similar foundation species subjected to environmental stress.

The adult population dynamics and ecology of Gyrinus substriatus in one of several small water—filled quarries on an island in southeast Sweden were studied over several years. The entire population was censused with a mean interval of 14 d. Individuals were marked to indicate the first date they were observed and tenerals were distinguished from older beetles. These methods provided records of longevity, mortality, and reproductive success. The maximum population varied by a factor of 2.5 over six normal years. During two consecutive years the population went extinct, when low water level plus overgrowth of the perennial macrophyte Juncus bulbosus caused greatly reduced water surface area. During the extinction years, the beetles were attacked by a parasitic water mite (Eylais sp.) and several dead beetles were found. There was a high late spring—early summer immigration, but no exchanges could be shown between adjacent populations. This, in combination with records in the quarry ofspecies that did not breed in the vicinity of the surveyed locality, suggests regular long—distance dispersal. My data suggest that the size of the new generation generally is not proportional to the size of the locally hibernated fraction. This may or may not indicate some sort of density—dependent regulating factors. With the high winter mortality, the between—year population fluctuations were much affected by large—scale recruitment of spring colonizers from remote localities. G. substriatus had a long life—span and 4—13% of the immigrants overwintered 2 yr as adults. But winter mortality was severe; 47—86% of late—autumn beetles died by the following spring. Immigrated 1—yr adults had a higher survival rate during the 1st mo than tenerals. Survival of tenerals was related to season; early emergers had much higher survival than very late emergers. Each female produced between 11 and 16 adult offspring per year. The experimentally introduced G. aeratus, with which G. substiatus ofen co—occurs in mainland localities, had high adult survival. It had higher survival during a year when all G. substriatus were removed experimentally, perhaps indicating competitive release. This is supported by preliminary findings of identical prey choice for adults ot these two species. G. aeratus reproduced in both years when it was introduced, but only °30% as successfully as did G. substriatus. The young of G. aeratus were unaffected by the presence of G. substriatus. High death rate of teneral G. aeratus led to extinction °1 yr after each experimental introduction. The local but scarce G. natator did not increase after introduction.

Flights (movements from place to place) were studied in 11 species of Lepidoptera which differed in size, mate—locating behavior, food plants of larvae and adults, oviposition behavior, population size, areal extent of populations, and habitat features possibly limiting flights. and population size were quantified. Mate—locating behavior influences the difference in flights of ♂ ♂ and ♀ ♀; males tend to remain in sites favorable for mating, while ♀ ♀ tend to disperse more than ♂ ♂. Multibrood polyphagous species feeding on early successional plants had the farthest flights. Conversely, single—brood species feeding as larvae on perennial trees or shrubs had short flights. There is apparently a genetic component to flight distances; taxonomically similar species had similar distances, flights of several species were similar in successive years, and flights of reared individuals were similar to those of native individual. Two species in different superfamilies occupying the same habitat convergently developed similar flight distances. A migratory species had more unidirectional flight than the nonmigrants. There is a strong positive correlation between the size of the area in which the majority of individuals of a population are concentrated and flight distances for the 11 species. No correlation exists with population size, but mean densities are inversely correlated with flight distances. The significance of these correlations is discussed. It ispostulated that heritable behavioral mechanisms and local extinction influence the relationships between flights, area, and density.

Four species of heteromyid rodents were studied in 5 communities in the Lower Sonoran Desert to examine differences in resource utilization by heteromyids in terms of: (1) collection of seed species, seed size and seed type (shrub—tree, forb and grass) and (2) association with amount of perennial cover of habitat. These rodents were not allocating resources by seed species or seed size. These rodents were collecting seeds to some degree by seed type, however, this was related to foraging in particular microhabitats. We found no correlation between numbers of each species of seed in cheek pouch contents and energy content per seed species. These heteromyids were opportunistic in collecting seeds as shown by considerable variation in seed species collected by populations and species among habitatas and seasonal variation in seed species collected. These heteromyids appear to partition seed resources by specializing on particular microhabitats. If these heteromyids use microhabitats optimally as granivores, they cannot allocate resources by seed species, size or type without risking local extinction due to the high variable production of specific seed species and sizes in the desert. The regular distribution of body sizes among heteromyids in a community appears to be a consequence of exploiting particular microhabitats in the desert community rather than allocating resources by seed—size selection.

Sporophytes of the sea palm Postelsia normally occur on a variety of inter- tidal surfaces. An experimental technique was developed permitting spores to be sown onto natural substrata whose relative suitabilities were evaluated using three criteria. All major surface categories (bare rock, animal, plant) were equivalent if judged by the appearance the following spring of a descendant sporophyte. However, the proportion of these plants surviving to maturity and continuing into the following year was highly substratum de- pendent. Bare rock surfaces are the most suitable for Postelsia as judged by probability of successful occupancy, sporophyte density, and inter-annual persistence. Single fertile plants can establish or maintain a population. The probability of local extinction is a function of population size: only 36% of small populations (1-30 plants) tended to continue to the next year, whereas all those > 120 individuals did. Postelsia seems incapable of persisting in the presence of the turf-like algae Corallina or Halosaccion. Thus sea palms may require the presence of competitively superior mussels: mussels out- compete algae of low stature; waves remove mussels, thereby generating Postelsia's most suitable substratum.

Localized disturbances transform most assemblages of sessile organisms into mosaics of patches differing in characteristics such as size and age (time since last disturbed). This mosaic nature of natural communities is especially evident on exposed intertidal shores along the northwest coast of North America, where the competitively dominant mussel, Mytilus californianus, occupies much of the space at mid-tidal levels. Nearly continuous beds of this species are interrupted by patches of open space generated mainly by the shearing forces of winter storm waves. These patches serve as foci for the recruitment, growth, and reproduction of many competitively inferior, "fugitive" species, including both algae and sessile invertebrates. These species are doomed to local extinction as the lateral encroachment of adult mussels closes the patch and excludes them from the area. This study examined the dynamics of algal succession within experimental patches cleared in mussel beds. In particular, two potentially important sources of variation in successional dynamics were investigated: (1) the size of the patch when first created, and (2) the location of the patch with respect to potential sources of propagules. The size of a cleared patch was found to influence strongly the course of algal succession. This effect was largely indirect, resulting from an interaction between patch size and grazing intensity. Small patches support higher densities of grazers, especially limpets, than do large patches. As a consequence, the assemblages of algae that develop within small and large patches differ markedly. The assemblage in small patches includes grazer-resistant but apparently competitively inferior species, whereas that in large patches is composed of grazer-vulnerable but competitively superior species. Small patches appear to serve as refuges from competition for grazer-resistant species. Recruitment was variable among the experimental patches. Percent cover of several species was found to be highly correlated with the cover of epizoic nonspecific adults within 1 m of the edge of the patch. This result suggests that a number of the species inhabiting patches within mussel beds may disperse their propagules over relatively short distances. For such species, patch dispersion may influence the regional dynamics of their populations.

We examine the qualitative behavior of differential equations for the proportion of insular patches of each of two kinds of habitat occupied by each of two species with characteristics rates of migration between patches and of local extinction within a habitat. Certain migration and extinction rates result in stable coexistence, even of closely similar species; others lead to competitive exclusion, even when each species is competitively superior in one kind of habitat. For a community of many species in many habitats, we surmise qualitative limits to the subdivision of resources, and alternative stable communities. Our results extend to species that live in successional or ephemeral habitats. We therefore conjecture equilibrial theories for the number of patches of habitat occupied by insular species, fugitive plants and invertebrates, or infesting parasites.

Newly created oceanic islands slowly accumulate species. In contrast, habitat islands created by rising water level or by human reduction of the area of a habitat (for example, by deforestation) support a full complement of local species at their formation. With time these islands lose species, a phenomenon called relaxation (Diamond 1972, Wilcox 1980). Several explanations may account for relaxation in island relicts. (1) As island size decreases, diversity of habitats present decreases. Species dependent entirely or in part on lost habitats disappear ("habitat diversity hypothesis;" Able and Connor 1979, Wilcox 1980). (2) Islands passively intercept dispersers in proportion to their area. As island size declines, recolonization probabilities decline. Thus, species lost from a small land bridge island are less likely to recolonize ("disperser intercept hypothesis;" Able and Connor 1979, Martin 1980, Wilcox 1980). (3) Smaller islands have more species with low population sizes, a condition which predisposes such species for extinction from random causes ("rarity hypothesis"; Willis 1974, Terborgh and Winter 1980). Corollaries of the "rarity hypothesis" relate bird size and ecological specialization to extinction probability (Willis 1974). (4) On theoretical grounds, species unable to maintain stable populations in the face of environmental vagaries (species with high "population variability") may be prone to local extinction (Leigh 1975, 1981). Terborgh and Winter (1980) suggested from qualitative considerations that species dependent on resources likely to be variable and/or patchy in their distribution (especially frugivores and nectarivores) are prone to extinction. Tracking of those resources in space results in locally variable population sizes for such species. To my knowledge, no quantitative field data are available to test the "population variability hypothesis." To what extent do these hypotheses account for loss of species on Barro Colorado Island (BCI), Panama, a 14.8-km2 land bridge island formed when the Panama Canal was created by damming the Chagras River early in the 20th century? I have used two approaches to explore that question. First, I have examined in detail the avifauna of BCI and its presumed source fauna in the adjacent mainland forest of the Pipeline Road region of Parque Nacional Soberania. Soberania is a 22 000-ha tract of lowland forest on the mainland adjacent to BCI. At its nearest point, BCI is ;200 m from a peninsular extension of that forest, although less-disturbed forest is not so close. As many as 5060 species of forest birds are missing from BCI (Karr 1982), well above earlier conclusions that 15-18 forest species are extinct on BCI (Terborgh 1974, Willis 1974, Wilson and Willis 1975, Willis and Eisenmann 1979). The approach used in earlier estimates of avian extinctions on BCI led to underestimation of extinction rates. Further, the high extinction rate seems to be due to the restricted habitat mosaic of BCI relative to that found on a similar-sized mainland area. Undergrowth and ground species and species associated with foothill forest are especially prone to extinction (Karr 1982). Second, I have conducted a long-term study of undergrowth species in mainland forest near BCI, with the objective of identifying population attributes of species prone to extinction. I report those findings here. Birds of forest undergrowth have been sampled periodically since 1968 with mist nets in an area of 2 km2 along the Pipeline Road (centered at Limbo Hunt Club [LHC]) and 9 km east of BCI. A total of 4722 captures of 114 species of birds is available in samples collected from August 1968 through March 1982. The methodology for this study was operation of 12-14 mist nets in the undergrowth of forest for periods of 3-6 d (see Karr [1979, 1981] for details of methodology). All data were collected by me except for 10 monthly samples (826 captures) from 1977 to 1978 collected by Schemske and Brokaw (1981). Net shyness sometimes occurs in areas that are regularly sampled with mist nets. However, sampling in recent years has been limited to two 3-6 d periods each year, and.there is no sign of net shyness with this low sampling frequency. Indeed, 5070W of captures are of birds banded during earlier sampling periods. In this paper I concentrate on two of the four hypotheses described above: rarity and population variability. I do not consider the disperser intercept hypothesis since it is incompatible with the empirical observation that the undergrowth forest birds discussed in detail here are very reluctant to cross water gaps. The habitat diversity hypothesis will be discussed only briefly here as it relates to population variability since it is the primary focus of another paper (Karr 1982). I assume that the relative abundances of species today in the vicinity of Limbo Hunt Club are similar to those of the bird community of BCI at the time BCI became an island.

to environmental extremes of drought and flash floods. 2. The natural life span for most Rhinichthys oscults in the Chiricahua Mountains is less than 3 years. None was found to reach 4 years. 3. High mortality rates during summer drought must be interpreted, in part, as an acceleration of death among older fish since most fish are expected to die before they are 3 years old. Among younger fish, drought directly elevates the mortality rate. 4. Mortalities during a drought are caused directly by the disappearance of water and indirectly by starvation of the fish which are crowded into reduced habitat with inadequate food. 5. Flash floods are an important cause of mortality among fish of the year if they occur while the fish are very small. The greatest potential loss of fish of the year occurs in late summer when the initial flash flood induces major reproduction and is then followed by another flash flood. 6. Flash floods are not a significant cause of mortality among older fish. 7. Temporary waters may persist in some sections through a period of wet years. The populations of fish inhabiting them are derived from upstream sections during a flash flood. All fish that were located in downstream temporary sections in the spring of 1960, were one-year-olds. 8. Temperatures do not rise to lethal levels in any of the flowing streams. In exposed shallow isolated pools, the temperatures may become lethal for oldcr fish, but not for fish of the year. 9. Predators play a minor role. Only one, the garter snake, Thamnophtis cyrtopsis, is common and apparently important. 10. Although fish populations are greatly reduced by a year of drought, there is no immediate threat of their local extinction.