Spatial ecology of host–parasitoid interactions: a threatened butterfly and its specialised parasitoid

Habitat conservation for threatened temperate insect species is often guided by one of two paradigms: a metapopulation approach focusing on patch area, isolation and number; or a habitat approach focusing on maintaining high quality habitat for the focal species. Recent research has identified the additive and interacting importance of both approaches for maintaining populations of threatened butterflies. For specialised host-parasitoid interactions, understanding the consequences of habitat characteristics for the interacting species is important, because (1) specialised parasitoids are particularly vulnerable to the consequences of fragmentation, and (2) altered interaction frequencies resulting from changes to habitat management or the spatial configuration of habitat are likely to have consequences for host dynamics. The spatial ecology of Cotesia bignellii, a specialist parasitoid of the threatened butterfly Euphydryas aurinia, was investigated at two spatial scales: within habitat patches (at the scale of individual aggregations of larvae, or ‘webs’) and among habitat patches (the scale of local populations). Parasitism rates were investigated in relation to larval web size, vegetation sward height and host density. Within patches, the probability of a larval webs being parasitized increased significantly with increasing number of larvae in the web, and parasitism rates increased significantly with increasing web isolation. The proportion of webs parasitized was significantly and negatively correlated with cluster density. Among habitat patches the proportion of parasitized webs decreased as cluster density increased. Clusters with a high proportion of larval webs parasitized tended to have lower parasitism rates per larval web. These results support the call for relatively large and continuous habitat patches to maintain stable parasitoid and host populations. Conservation efforts directed towards maintenance of high host plant density could allow E. aurinia to reduce parasitism risk, while providing C. bignellii with sufficient larval webs to allow population persistence.     

The dragonfly delusion: why it is essential to sample exuviae to avoid biased surveys

Odonate populations and species numbers are declining globally. Successful conservation requires sound assessments of both odonate distributions and habitat requirements. Odonates have aquatic (larval) and terrestrial (adult) stages, but most surveys that are used to inform conservation managers are undertaken of the adult stage. This study investigates whether this bias towards adult records in odonate recording is misinterpreting the environmental quality of sites. The habitat focus is farmland ponds, a key feature of agricultural landscapes. We tested whether or not, adult, larval and exuvial surveys lead to similar conclusions on species richness and hence on pond quality. Results showed that pond surveys based upon larvae and exuviae are equally suitable for the reliable assessment of presence/absence of odonates, but that adult surveys are not interchangeable with surveys of larvae/exuviae. Larvae were also found at ponds with no emerging individuals due to changes in habitat quality, therefore presence of exuviae remains the only proof of life-cycle completion at a site. Ovipositing females were recorded at all ponds where exuviae were totally absent hence adult surveys over-estimate pond quality and low-quality ponds are functioning as ecological traps. Highly mobile and generalist species were recorded at more locations than other species. Adult surveys also bias recording towards genera, species and populations with non-territorial mate-location strategies. Odonate biodiversity monitoring would benefit from applying the best survey method (exuviae) to avoid wasting valuable financial resources while providing unbiased data, necessary to achieve conservation objectives.     

Functional habitat area as a reliable proxy for population size: case study using two butterfly species of conservation concern

Accurate estimates of population size are essential for effective conservation and restoration management of threatened species. Nevertheless, reliable methods to estimate population size, such as mark-release-recapture studies (MRR), are time and labour consuming and may generate negative impact(s) on both the habitats and organisms studied. This may complicate their use if several sites need to be studied concurrently. Consequently, there is a strong interest to develop reliable proxies of population size, e.g., to be used in Population Viability Analysis. Habitat area has often been used as an obvious proxy. For butterflies, many studies focused on the area of host plant patches, but resource-based definition of the habitat (i.e., the area containing the different ecological resources and conditions needed by the individuals) has recently gained much attention. Using two peat bog butterflies, we tested the reliability of these two measures of habitat area as proxies for population size by (1) predicting population sizes based on the product of larval habitat area by the number of emerged butterflies per spatial unit of habitat (eliminated by ground cover traps) and (2) comparing these predictions to accurate population size estimates inferred from MRR studies. Results on both species showed that: (1) adult population size was strongly related to larval habitat availability and quality when habitat was accurately defined according to functional resources, (2) resources other than the host plant have to be included in the habitat definition, (3) after careful control of its similarity, the resource-based habitat delineation can be reasonably well transferred among populations of the same species in a wider region.     

How limited is dispersal in the rare beetle, Cryptocephalus decemmaculatus (Chrysomelidae, Cryptocephalinae)?

As suitable habitat becomes increasingly fragmented the ability of a species to reach new areas may often dictate whether populations persist or perish. An understanding of dispersal ability is a prerequisite for informed management decisions. This is particularly true for species that have become restricted to one or a small number of sites, as is the case for several species of Cryptocephalus beetle in the UK. Using mark–release–recapture, host-plant marking and direct observations we investigated population size, movement through suitable habitat and the flight behaviour of Cryptocephalus decemmaculatus. In the UK this species is known from only two sites and it is listed as a priority species on the UK Biodiversity Action Plan. At the time of the study, the one known English site for this species supported a population of ~500 individuals. The adult beetles move through their habitat using suitable host-plants as ‘stepping stones’. There appears to be no interchange of adults between sub-populations that are separated by small areas of unsuitable habitat (e.g. small tracts of woodland or areas devoid of scrub), a result reinforced by studies of genetic differentiation between sub-populations. The small population size of this beetle and its association with, mid-successional habitats makes it vulnerable to local extinction. Furthermore, its limited dispersal ability means that other habitat patches are unlikely ever to be colonised naturally.     

Species‐habitat relationships and ecological correlates of butterfly abundance in a transformed tropical landscape

Tropical butterfly conservation strategies often focus on total and/or common species richness to assess the conservation value of a patch or habitat. However, such a strategy overlooks the unique dynamics of rare species. We evaluated the species‐habitat relationships of 209 common, intermediate, and rare butterfly species (including morphospecies) across four habitat types (mature, degraded, or fragmented forest, and urban parks) and two patch sizes (<400 ha, ≥400 ha) in Singapore. Common species richness was consistent across habitat types. Intermediate species richness declined by more than 50 percent in urban parks (relative to all forest habitats), and rare species richness was reduced by 50 percent in degraded and fragmented forest and by 90 percent in urban parks (relative to mature forest). Large patches had comparable overall richness to small patches, but they supported more rare species and three times as many habitat‐restricted species over a similar area. Importantly, a number of rare species were confined to single small patches. Mixed‐effects regression models were constructed to identify habitat and ecological/life history variables associated with butterfly abundance. These models revealed that species with greater habitat specialization, rare larval host plants, few larval host plant genera, and narrow global geographic ranges were more likely to be rare species. Overall, these results demonstrate that the richness of habitat‐restricted and rare species do not follow the same spatial distribution patterns as common species. Therefore, while conserving mature forests is key, effective butterfly conservation in a transformed landscape should take into account rare and habitat‐restricted species.     

Assessing Extinction Risk in Small Metapopulations of Golden‐headed Lion Tamarins (Leontopithecus chrysomelas) in Bahia State,Brazil

Golden‐headed lion tamarins (GHLTs; Leontopithecus chrysomelas) are endangered primates endemic to the Brazilian Atlantic Forest, where loss of forest and its connectivity threaten species survival. Understanding the role of habitat availability and configuration on population declines is critical for guiding proactive conservation for this, and other, endangered species. We conducted population viability analysis to assess vulnerability of ten GHLT metapopulations to habitat loss and small population size. Seven metapopulations had a low risk of extirpation (or local extinction) over the next 100 years assuming no further forest loss, and even small populations could persist with immediate protection. Three metapopulations had a moderate/high risk of extirpation, suggesting extinction debt may be evident in parts of the species’ range. When deforestation was assumed to continue at current rates, extirpation risk significantly increased while abundance and genetic diversity decreased for all metapopulations. Extirpation risk was significantly negatively correlated with the size of the largest patch available to metapopulations, underscoring the importance of large habitat patches for species persistence. Finally, we conducted sensitivity analysis using logistic regression, and our results showed that local extinction risk was sensitive to percentage of females breeding, adult female mortality, and dispersal rate and survival; conservation or research programs that target these aspects of the species’ biology/ecology could have a disproportionately important impact on species survival. We stress that efforts to protect populations and tracts of habitat of sufficient size throughout the species’ distribution will be important in the near‐term to protect the species from continuing decline and extinction.     

Settlement in different-sized patches by the gregarious intertidal barnacle Chamaesipho tasmanica Foster and Anderson in New South Wales

Most models on settlement of open marine invertebrate populations are based on space-limitation. These models, however, do not recognise that free space may not drive the demography of populations when larval numbers are small or when larval supply varies along a gradient in the habitat. They also do not incorporate the effects of larval choice when settling. It has been hypothesised that, in gregarious barnacles, the effects of adult conspecifics, rather than available free space, may play a primary role in settlement. That is, cues from adults along perimeters of patches, rather than space available, may enhance colonisation. This study therefore aimed to distinguish between these separate influences on populations of Chamaesipho tasmanica, a gregarious barnacle characterised by relatively few larvae arriving to settle each year. Patches of 6, 3 and 1.5 cm diameter were cleared within aggregations of barnacles at three heights (Low, Mid, Upper) of Chamaesipho's distribution at two sites and during 2 years of settlement. Total numbers of settlers in each year were manipulated to determine the separate influences on settlement due to availability of substratum or the effects of conspecific adults. To test for the effects of available free space, numbers of settlers per unit area were analysed. To test for gregarious effects due to the presence of adults, numbers of settlers per unit perimeter were analysed. While available substratum was found not to affect settlement of this barnacle, gregarious settlement in response to adults at perimeters of patches was thought to be confounded by differential larval supply and differential conspecific cues among heights on the shore. Results from this study therefore have important implications for survival of gregarious populations following disturbances, especially in species where larval supply is poor.     

Marine reserve design theory for species with ontogenetic migration

Models for marine reserve design have been developed primarily with ‘reef fish’ life histories in mind: sedentary adults in patches connected by larval dispersal. However, many fished species undertake ontogenetic migrations, such as from nursery grounds to adult spawning habitats, and current theory does not fully address the range of reserve options posed by that situation. I modelled a generic species with ontogenetic migration to investigate the possible benefits of reserves under three alternative scenarios. First, the fishery targets adult habitat, and reserves can sustain yields under high exploitation, unless habitat patches are well connected. Second, the fishery targets the nursery, and reserves are highly effective, regardless of connectivity patterns. Third, the fishery targets both habitats, and reserves only succeed if paired on adjacent, well-connected nursery and adult patches. In all cases, reserves can buffer populations against overexploitation but would not enhance fishery yield beyond that achievable by management without reserves. These results summarize the general situations in which management using reserves could be useful for ontogenetically migrating species, and the type of connectivity data needed to inform reserve design.     

Use of simple remote sensing tools to expedite surveys for rare tiger beetles (Insecta: Coleoptera: Cicindelidae)

Tiger beetles (Insecta: Coleoptera: Cicindelidae) often occupy small patches of suitable habitat in otherwise unsuitable landscapes. Such patches are easily overlooked, which may lead to underestimates of both the number of occurrences and the overall population size. In this study, simple World Wide Web-based tools (Google Earth and Microsoft Terraserver) were used to search high-resolution satellite imagery for patches of suitable habitat for globally and regionally rare tiger beetles on a 3,278 ha wildlife refuge in Maryland, USA. This tract is largely forested but contains scattered small open areas of sand and clay soils that are potential habitat for tiger beetles of conservation concern. Visual inspection of remotely sensed imagery resulted in the identification of 19 potential habitat patches, 15 of which yielded tiger beetle populations when surveyed on the ground. The number of species of tiger beetles recorded from this tract was increased from 3 to 8 and two new sites were discovered for the state sensitive species Cicindela scutellaris rugifrons Dejean. In addition, a small population of C. splendida Hentz was discovered, a species last reported from Maryland in 1948. The technique described here shows great promise for locating small patches of potential tiger beetle habitat in otherwise unsuitable landscapes.     

Dispersal-induced desynchronization: from metapopulations to metacommunities

The conceptualization of fragmented populations in terms of metapopulation theory has become standard over the last three decades. It is well known that increases in between‐patch migration rates cause more synchronous population fluctuations and that this coherence increases the risk of global metapopulation extinction. Because species’ migration rates and the probability of individuals surviving migration events depend on the effective distance between patches, the benefit of improving conservation corridors or the matrix between habitat patches has been questioned. As populations occur in the context of larger communities, moving from a metapopulation to a metacommunity model framework is a natural extension to address the generality of these conclusions. We show how considering a metacommunity can modify the conclusion that decreasing the effective distance between habitat patches (via improving matrix quality or other measures) necessarily increases the degree of metapopulation synchrony. We show that decreases in effective between‐patch distance may deter population synchrony because of the simultaneous effect this change has on the migration patterns of other species. These results indicate that species interactions need to be considered when the effect of conservation measures on population synchrony, and ultimately persistence, is addressed.     

Does variation in patch use among butterfly species contribute to nestedness at fine spatial scales?

A nested pattern occurs whenever the species observed in depauperate habitat patches are a subset of those found in more species‐rich patches. Ecologists have documented many instances of nestedness caused by population‐level processes such as colonization and extinction at biogeographic scales. However, few researchers have examined whether nestedness may exist at fine scales due to the ways in which individual organisms discriminate among potential habitat patches. In 1999, we experimentally fragmented an old‐field habitat into patches of varying size to test whether nestedness could exist on a fine spatial scale. Five treatments of differing patch size were replicated five times in a Latin square design by selectively mowing 15×15 m² plots within an old‐field (patch areas: 225, 180, 135, 90, and 45 m²). Specifically, we tested whether butterflies foraging within a network of patches differing in area conformed to a nested subset structure. We also classified species according to (1) their flight height while foraging (high or low), and (2) their adult habitat breadth (ubiquitous, general, or restricted) to determine whether nestedness could be explained by difference in species’ tendency to discriminate among patches differing in area.
We found significant evidence that a community of foraging Lepidoptera conformed to a nested subset structure based on the difference between the observed nestedness within the butterfly community and the nestedness obtained from randomly generated species presence/absence matrices. Poisson regression analyses demonstrated that high‐flying, habitat‐restricted species avoided the smallest patches (90 and 45 m²) in favor of larger remnants, whereas low‐flying, habitat generalists used all patch sizes. Thus, our study is one of the first to demonstrate that nestedness among species subsets can be observed at fine spatial scales (within a single 1.5 hectare field) and may be maintained by species behavioral differences: discriminating species (i.e. high‐flying, habitat restricted) avoided the smallest patches, and less discriminating species (i.e. low‐flying, ubiquitous) were distributed throughout the field without regard to patch size. Our results also suggest that nestedness should be viewed as yet another scalar pattern in ecology, generated by variation in patch use by individuals at fine‐scales as well as the more traditionally invoked processes of extinction and colonization of species at broad‐scales.     

Extinction-colonization dynamics and host-plant choice in butterfly metapopulations

Species living in highly fragmented landscapes often occur as metapopulations with frequent population turnover. Turnover rate is known to depend on ecological factors, such as population size and connectivity, but it may also be influenced by the phenotypic and genotypic composition of populations. The Glanville fritillary butterfly (Melitaea cinxia) in Finland uses two host-plant species that vary in their relative abundances among distinct habitat patches (dry meadows) in a large network of approximately 1,700 patches. We found no effect of host species use on local extinction. In contrast, population establishment was strongly influenced by the match between the host species composition of an empty habitat patch and the relative host use by larvae in previous years in the habitat patches that were well connected to the target patch. This "colonization effect" could be due to spatially variable plant acceptability or resistance or to spatially variable insect oviposition preference or larval performance. We show that spatial variation in adult oviposition preference occurs at the relevant spatial scale and that the other possible causes of the colonization effect can be discounted. We conclude that the colonization effect is generated by host preference influencing the movement patterns of ovipositing females. Migrant females with dissimilar host preferences have different perceptions of relative patch quality, which influences their likelihood of colonizing patches with particular host composition.     

Effects of meadow suitability on female behaviour in the alpine butterfly Parnassius smintheus

Abstract  1. Many butterfly populations persist in networks of naturally fragmented habitat patches. Movement and reproductive decisions made by adult females are critical to the persistence of these populations because colonisation of extinct habitat patches in the network requires emigration of fecund adult females from their natal meadow and their subsequent establishment in the extinct patch.
2. Movement and oviposition behaviours of mated Parnassius smintheus females released in suitable meadows (a good- and a poor-quality meadow) and an unsuitable meadow were compared, to determine whether adult females consider meadow suitability for their offspring despite frequent oviposition events off the larval host plant.
3. Bootstrap and correlated random walk analyses of female step lengths and turn angles demonstrated that females flew more randomly in the unsuitable meadow than in the suitable meadows. Although females tended to turn the sharpest angle between landing sites in the good-quality meadow, and fly the smallest distance between landing sites and displace the smallest distance from the release site in the suitable meadows, no significant differences were detected in turn angle, step length, and dispersal rates between suitable and unsuitable meadows.
4. Results from female flight observations and a caged oviposition study suggest that females lay significantly more eggs in suitable habitat than in unsuitable habitat despite not ovipositing on the host plant, and support the above findings.
5. Movement and oviposition behaviours of adult female P. smintheus promote their retention within meadows that can support their offspring.     

Host-plant patch qualities and presence of a likely competitor species affect the distribution and abundance of a rare British moth, <Emphasis Type="Italic">Cucullia lychnitis</Emphasis>

Understanding the environmental factors that affect the distribution and abundance of rare and threatened species can help identify priority sites for conservation action such as habitat management. The moth Cucullia lychnitis is identified in UK legislation as a priority species for conservation. Its conspicuous larvae feed on a patchily distributed host plant Verbascum nigrum, but are entirely absent from some host plant patches within the species’ range. Environmental variability among patches was investigated to understand factors affecting (i) patch occupancy by C. lychnitis and (ii) abundance of C. lychnitis, where present. Occupancy of individual V. nigrum plants within occupied patches was also investigated. The likelihood of patch occupancy increased with patch size, and decreased with patch isolation. Abundance of C. lychnitis was negatively correlated with patch isolation and with abundance of the weevil Cionus nigritarsis, which exploits the same host plant and may be a competitor. Within occupied patches, larvae of C. lychnitis were significantly more likely to be found on taller plants with more flower spikes; such plants are typical of established rather than newly-created patches. These results may help to guide efforts to conserve C. lychnitis, highlighting the importance of habitat connectivity and indicating potential management actions to promote features positively associated with larval incidence and abundance.     

Can site and landscape‐scale environmental attributes buffer bird populations against weather events?

Projected impacts of climate change on the populations and distributions of species pose a challenge for conservationists. In response, a number of adaptation strategies to enable species to persist in a changing climate have been proposed. Management to maximise the quality of habitat at existing sites may reduce the magnitude or frequency of climate‐driven population declines. In addition large‐scale management of landscapes could potentially improve the resilience of populations by facilitating inter‐population movements. A reduction in the obstacles to species’ range expansion, may also allow species to track changing conditions better through shifts to new locations, either regionally or locally. However, despite a strong theoretical base, there is limited empirical evidence to support these management interventions. This makes it difficult for conservationists to decide on the most appropriate strategy for different circumstances. Here extensive data from long‐term monitoring of woodland birds at individual sites are used to examine the two‐way interactions between habitat and both weather and population count in the previous year. This tests the extent to which site‐scale and landscape‐scale habitat attributes may buffer populations against variation in winter weather (a key driver of woodland bird population size) and facilitate subsequent population growth. Our results provide some support for the prediction that landscape‐scale attributes (patch isolation and area of woodland habitat) may influence the ability of some woodland bird species to withstand weather‐mediated population declines. These effects were most apparent among generalist woodland species. There was also evidence that several, primarily specialist, woodland species are more likely to increase following population decline where there is more woodland at both site and landscape scales. These results provide empirical support for the concept that landscape‐scale conservation efforts may make the populations of some woodland bird species more resilient to climate change. However in isolation, management is unlikely to provide a universal benefit to all species.     

Review of the conservation status of the Atlas Moth, <Emphasis Type="Italic">Attacus wardi</Emphasis> Rothschild, 1910 (Lepidoptera: Saturniidae) from Australia

The conservation status of Attacus wardi, a large iconic moth endemic to north-western Australia, is reviewed based on new data. Available evidence on the spatial distribution, critical habitat and threatening processes suggests the species qualifies as threatened according to IUCN Red List Criteria, and that its conservation status nationally should be revised from Endangered to Vulnerable. The species depends on relatively large patches of wet and dry coastal tropical monsoon forest, and it has the potential to be an important flagship species for the conservation of these ecological communities. Further studies are needed to determine minimum patch size and spatial connectivity among patches to support viable populations of the moth.     

Patch occupancy in the endangered butterfly Lycaena helle in a fragmented landscape: effects of habitat quality, patch size and isolation

While there is agreement that both habitat quality and habitat network characteristics (such as patch size and isolation) contribute to the occupancy of patches by any given species, the relative importance of these factors is under debate. This issue is of fundamental ecological importance, and moreover of special concern for conservation biologists aiming at preserving endangered species. Against this background we investigated patch occupancy in the violet copper Lycaena helle, one of the rarest butterfly species in Central Europe, in the Westerwald area (Rhineland-Palatinate, Western Germany). Occupied (n = 102) differed from vacant (n = 128) patches in altitude, size, connectivity, availability of wind shelter, in the abundance of the larval host-plant, in the abundance of a grass species indicating favorable habitat conditions and in the abundance of nitrophilous plants. Overall, patch occupancy was primarily determined by patch size, connectivity and the abundance of the larval host plant, while all other parameters of habitat quality were of subordinate importance. Therefore, our findings suggest that even for extremely sedentary species such as L. helle habitat networks are decisive and—next to the preservation of habitat quality—need to be an integral part of any conservation management for this species.     

Butterflies on the brink: habitat requirements for declining populations of the marsh fritillary (Euphydryas aurinia) in SW England

1. The marsh fritillary Euphydryas aurinia is one of our most endangered butterflies, and the only to be protected under European legislation as well as British. It persists in fragile subpopulations threatened by habitat fragmentation and degradation. 2. A combination of swaling and cattle grazing are accepted to be best practice for managing wet, unimproved grasslands??the favoured habitat for E. aurinia in Cornwall. These two well-endorsed methods of management were used to increase and improve the quality of habitat for E. aurinia over a 5 years period, 2004?C2008, at a stronghold network of habitat patches in mid Cornwall, south-west England. 3. Analyses of adult and larval densities over 5 years in fifty-four transects across nine sites found E. aurinia to favour habitat patches with higher densities of the larval food plant (Devil??s-bit scabious Succisa pratensis), higher sward height in autumn, and intermediate optimum levels of stock grazing. 4. Main findings indicated most sites experienced significant declines in numbers. Unfavourable weather in the last 2 years of monitoring was likely to have had a significant impact on the response of individual subpopulations to habitat management though poor recovery rates may also reflect a time-lag in colonisation events after habitat improvement has occurred. 5. Habitat management produced an improvement, albeit an inconsistent improvement in habitat variables across patches??S. pratensis shows a clear recovery at some sites. Autumn sward height increased significantly at one site, and a quadratic relationship between stock grazing and important habitat variables has been found which will aid further improvement over all sites for the long term persistence of E. aurinia.     

Species’ surrogacy for conservation planning: caveats from comparing the response of three arboreal rodents to habitat loss and fragmentation

The use of surrogate species in conservation planning has been applied with disappointing results on relatively large sets of species. It could still prove useful for optimizing conservation efforts when considering a small set of species with similar ecological requirements, however few field tests of this nature have been carried out. The aim of this research is to compare the response of three arboreal rodent species—the fat dormouse (Glis glis), the hazel dormouse (Muscardinus avellanarius) and the red squirrel (Sciurus vulgaris)—to habitat loss and fragmentation, with the aim of identifying priorities for conservation and evaluating possible optimization of conservation efforts under different scenarios: habitat restoration and selection of focal patches. We studied the distribution of the three species in a sample of patches in a highly fragmented landscape in central Italy, using a patch-landscape scale approach. The distribution was studied by using hair tubes, nestboxes and nocturnal surveys. The three species showed analogous responses to increasing isolation and decreasing size of habitat patches; what differed however, was the magnitude of responses. Our results show possible application of surrogacy within this restricted group of species, however several caveats arise depending on the conservation strategy and available funding. If habitat restoration is the objective, then the fat dormouse should be the target species for guiding size and isolation of patches. On the other hand, the magnitude of the differences and patch requirements for this species, question the feasibility of these conservation actions. If selection of focal patches for conservation is the objective then selecting the fat dormouse as a focal/umbrella species would overlook areas suitable for the other two species. Feasible optimisation of conservation efforts may be possible only between the red squirrel and the hazel dormouse.     

Chaotic dynamics may determine the effect of inter-patch migration on metapopulation survival

A simple, strategic model of a system of habitat fragments connected by conservation corridors is presented. The intrinsic dynamics of the population on each fragment are stochastic. In addition, at each generation there is a probability of a catastrophic event occurring which affects all the habitat fragments by greatly reducing the size of the population on each. Global extinction is considered to occur when all the populations simultaneously fall below a threshold value. If the intrinsic dynamics on each fragment are simple cycles or a stable equilibrium, then the addition of conservation corridors does not reduce the frequency of global extinction. This is because migration between fragments induces their populations to have values which are similar to each other. However, if the intrinsic population dynamics are chaotic then the probability of global extinction is greatly reduced by the introduction of conservation corridors. Although local extinction is likely, the chaos acts to oppose the synchronising effect of migration. Often a subset of the populations survive a catastrophe and can recolonize the other patches.