Pedro Beja

According to ecological theory, the coexistence of competitors in patchy environments may be facilitated by hierarchical spatial segregation along axes of environmental variation, but empirical evidence is limited. Cabrera and water voles show a metapopulation-like structure in Mediterranean farmland, where they are known to segregate along space, habitat, and time axes within habitat patches. Here, we assess whether segregation also occurs among and within landscapes, and how this is influenced by patch-network and matrix composition. We surveyed 75 landscapes, each covering 78 ha, where we mapped all habitat patches potentially suitable for Cabrera and water voles, and the area effectively occupied by each species (extent of occupancy). The relatively large water vole  tended to be the sole occupant of landscapes with high habitat amount but relatively low patch density (i.e., with a few large patches), and with a predominantly agricultural matrix, whereas landscapes with high patch density (i.e., many small patches) and low agricultural cover, tended to be occupied exclusively by the small Cabrera vole. The two species tended to co-occur in landscapes with intermediate patch-network and matrix characteristics, though their extents of occurrence were negatively correlated after controlling for environmental effects. In combination with our previous studies on the Cabrera-water vole system, these findings illustrated empirically the occurrence of hierarchical spatial segregation, ranging from within-patches to among-landscapes. Overall, our study suggests that recognizing the hierarchical nature of spatial segregation patterns and their major environmental drivers should enhance our understanding of species coexistence in patchy environments.

Species distribution models combining environmental and spatial components are increasingly used to understand and forecast species invasions. However, modelling distributions of invasive species inhabiting stream networks requires due consideration of their dendritic spatial structure, which may strongly constrain dispersal and colonization pathways. Here we evaluate the application of novel geostatistical tools to species distribution modelling in dendritic networks, using as case study two invasive crayfish (Procambarus clarkii and Pacifastacus leniusculus) in a Mediterranean watershed. Specifically, we used logistic mixed models to relate the probability of occurrence of each crayfish to environmental variables, while specifying three spatial autocorrelation components in random errors. These components described spatial dependencies between sites as a function of (1) straight-line distances (Euclidean model) between sites, (2) hydrologic (along the waterlines) distances between flow-connected sites (tail-up model), and (3) hydrologic distances irrespective of flow connection (tail-down model). We found a positive effect of stream order on P. clarkii, indicating an association with the lower and mid reaches of larger streams, while P. leniusculus was affected by an interaction between stream order and elevation, indicating an association with larger streams at higher altitude. For both species, models including environmental and spatial components far outperformed the pure environmental models, with the tail-up and the Euclidean components being the most important for P. clarkii and P. leniusculus, respectively. Overall, our study highlighted the value of geostatistical tools to model the distribution of riverine and aquatic invasive species, and stress the need to specify spatial dependencies representing the dendritic network structure of stream ecosystems.

ABSTRACT Aim Although the negative effects of habitat fragmentation have been widely documented at the landscape scale, much less is known about its impacts on species distributions at the biogeographical scale. We hypothesize that fragmentation influences the large-scale distribution of area- and edge-sensitive species by limiting their occurrence in regions with fragmented habitats, despite otherwise favourable environmental conditions. We test this hypothesis by assessing the interplay of climate and landscape factors influencing the distribution of the calandra lark, a grassland specialist that is highly sensitive to habitat fragmentation. Location Iberia Peninsula, Europe. Methods Ecological niche modelling was used to investigate the relative influence of climate/topography, landscape fragmentation and spatial structure on calandra lark distribution. Modelling assumed explicitly a hierarchically structured effect among explanatory variables, with climate/topography operating at broader spatial scales than landscape variables. An eigenvector-based spatial filtering approach was used to cancel bias introduced by spatial autocorrelation. The information theoretic approach was used in model selection, and variation partitioning was used to isolate the unique and shared effects of sets of explanatory variables. Results Climate and topography were the most influential variables shaping the distribution of calandra lark, but incorporating landscape metrics contributed significantly to model improvement. The probability of calandra lark occurrence increased with total habitat area and declined with the number of patches and edge density. Variation partitioning showed a strong overlap between variation explained by climate/topography and landscape variables. After accounting for spatial structure in species distribution, the explanatory power of environmental variables remained largely unchanged. Main conclusions We have shown here that landscape fragmentation can influence species distributions at the biogeographical scale. Incorporating fragmentation metrics into large-scale ecological niche models may contribute for a better understanding of mechanism driving species distributions and for improving predictive modelling of range shifts associated with land use and climate changes.

In farmland landscapes, amphibians often breed in artificial water bodies, which may offset at least partly the loss of natural wetlands associated with agricultural intensification. It is possible, however, that artificial habitats provide conditions for a minor proportion of the regional species pool, benefiting just a few widespread generalists. We assessed these alternative views by documenting the decline of temporary ponds in a Mediterranean farmland landscape between 1991 and 2009, and by estimating the occupancy of natural (temporary ponds and streams) and artificial water bodies (farm ponds, irrigation channels and drainage ditches) by amphibians across the breeding season (February–June 2010). We used hierarchical Bayesian dynamic multi-species occupancy modelling to control for differences in detectability across species, sampling occasions and habitat types. Over two decades the farmland landscape lost 56% of its temporary ponds, of which 89.3% were destroyed through agricultural activities such as cultivation, conversion to permanent farm ponds, and drainage. The survival rate of ponds was lowest within an irrigated perimeter, and there was no positive effect of protection by a natural park. Estimated species richness per site was at least twice as high in temporary ponds as in the other habitat types. From the 10 amphibian species recorded, seven had the highest occupancy in temporary ponds, and were absent or occurred rarely in artificial habitats. Only a single generalist species was widespread in farm ponds and irrigation channels. The results suggest that artificial water bodies are unlikely to sustain populations of most amphibian species in Mediterranean intensive farmland. Conservation efforts should be directed at protecting the temporary ponds still remaining within the farmed landscape, and at restoring or creating new temporary ponds where these have been lost during the last decades.

The bird collection of the Instituto de Investigação Cientítica Tropical (Lisbon, Portugal) holds 5598 preserved specimens (skins), mainly from Angola, Mozambique, Guinea-Bissau, São Tomé and Principe, and Cape Verde. The subset collection from Angola includes 1560 specimens, which were taxonomically revised and georeferenced for the publication of this data paper. The collection contains a total of 522 taxa, including 161 species and 361 subspecies. Two species are classified by the IUCN Red List as Endangered - the wattled crane (Grus carunculata) and the Gabela bush-shrike (Laniarius amboimensis) - and two are classified as vulnerable - African penguin (Spheniscus demersus) and the white-headed vulture (Trigonoceps occipitalis). The temporal span of the database ranges between 1943 and 1979, but 32% are from years 1958-1959, and 25% from years 1968-1969. The spatial coverage of the collection is uneven, with 2/3 of the records representing only four of the eighteen provinces of the country, namely Huíla, Moxico, Namibe and Cuanza Sul. It adds, however, valuable information for the Huíla area of the Angolan Scarp, which is probably a biodiversity hotspot of global conservation priority. Furthermore, this georeferenced database adds invaluable bird information to the GBIF network, for one of the countries with highest but less known biodiversity in Africa.

The metapopulation paradigm is central in ecology and conservation biology to understand the dynamics of spatially-structured populations in fragmented landscapes. Metapopulations are often studied using simulation modelling, and there is an increasing demand of user-friendly software tools to simulate metapopulation responses to environmental change. Here we describe the MetaLandSim R package, which integrates ideas from metapopulation and graph theories to simulate the dynamics of real and virtual metapopulations. The package offers tools to (i) estimate metapopulation parameters from empirical data, (ii) to predict variation in patch occupancy over time in static and dynamic landscapes, either real or virtual, and (iii) to quantify the patterns and speed of metapopulation expansion into empty landscapes. MetaLandSim thus provides detailed information on metapopulation processes, which can be easily combined with land use and climate change scenarios to predict metapopulation dynamics and range expansion for a variety of taxa and ecological systems.

Numerically, 42·7% (37·7% in biomass) of eagle prey comprised domestic species, about 70% of which were rural pigeons Columba livia and the remainder were racing pigeons Columba livia and domestic fowl Gallus gallus. Rabbits Oryctolagus cuniculus, red-legged ...

Although scrub cover is generally regarded as an important habitat component of the European wild rabbit Oryctolagus cuniculus, little is known about the species' responses to variation in scrub structure and composition. Such information is required for conservation ...

ABSTRACT Hunting and its associated management have significant costs and benefits for biodiversity conservation, which makes this socio-economic activity highly controversial at both international and regional levels. We investigated relationships between management for small game species (mainly Red-legged Partridges Alectoris rufa and rabbits Oryctolagus cuniculus ) and both abundance and richness of farmland and scrubland songbirds, raptors and ground-nesting birds, and on the abundance of three species of conservation concern (Little Bustard Tetrax tetrax , Eurasian Thick-knee Burhinus oedicnemus and Montagu’s Harrier Circus pygargus ) in southern Portugal farmland. We compared 12 game estates and 12 matching areas with similar sizes and land uses but no game management. Richness and abundance were estimated from fixed point counts, and were related to game regime (managed or unmanaged), habitat characteristics and census period. Our results showed that game management was associated, albeit weakly, with higher abundance of raptors and ground-nesting birds, but no relationship (either positive or negative) was found for other guilds and species. Habitat was generally the most important factor explaining bird species richness and abundance. Our results suggest possibilities for promoting management systems that could maximize both hunting sustainability and conservation value of managed areas, particularly when management helps to improve or maintain beneficial habitats or practices for farmland birds.

The bird collection of the Instituto de Investigação Cientítica Tropical (Lisbon, Portugal) holds 5598 preserved specimens (skins), mainly from Angola, Mozambique, Guinea-Bissau, São Tomé and Principe, and Cape Verde. The subset collection from Angola includes 1560 specimens, which were taxonomically revised and georeferenced for the publication of this data paper. The collection contains a total of 522 taxa, including 161 species and 361 subspecies. Two species are classified by the IUCN Red List as Endangered - the wattled crane (Grus carunculata) and the Gabela bush-shrike (Laniarius amboimensis) - and two are classified as vulnerable - African penguin (Spheniscus demersus) and the white-headed vulture (Trigonoceps occipitalis). The temporal span of the database ranges between 1943 and 1979, but 32% are from years 1958-1959, and 25% from years 1968-1969. The spatial coverage of the collection is uneven, with 2/3 of the records representing only four of the eighteen provinces of the country, namely Huíla, Moxico, Namibe and Cuanza Sul. It adds, however, valuable information for the Huíla area of the Angolan Scarp, which is probably a biodiversity hotspot of global conservation priority. Furthermore, this georeferenced database adds invaluable bird information to the GBIF network, for one of the countries with highest but less known biodiversity in Africa.

In Iberian cereal-steppes, decoupling of payments from current production levels through the Single Farm Payment raised concerns regarding the potential for land abandonment and replacement of sheep by cattle, with eventual negative consequences for declining grassland birds. This study addressed this issue by analysing the responses of five grassland bird species of conservation concern to spatial land use gradients, which are expected to reflect changes potentially associated with the CAP reform. Our results show that both habitat fragmentation and grazing regimes were major drivers of breeding bird densities, though responses to these factors were species-specific. Thekla larks were most abundant in landscapes with small grassland patches and high edge density, whereas calandra larks were abundant only in large expanses of continuous open farmland habitat. Little bustard and short-toed lark densities declined in highly fragmented landscapes, but they appeared to tolerate or even benefit from low to moderate levels of open habitat fragmentation. Corn buntings were little affected by landscape patterns. At the field scale, little bustard and corn bunting densities were highest in fields grazed by cattle, whereas short-toed larks were mostly associated with sheep pastures. Short-toed larks and Thekla larks were most abundant in old fallow fields where cattle was largely absent, whereas corn buntings showed the inverse pattern. These results confirm the view that the same agricultural policies may be favourable for some species of conservation concern but detrimental to others, and so they cannot be assumed to bring uniform conservation benefits.

Predicting the future distribution of a species under a variety of climate change scenarios is common practice. However, future projections of an Ecological Niche Model (ENM) calibrated under current conditions do not account for landscape characteristics or dispersal ability. Here we present MetaLandSim, a new R package, relying on the simulation of metapopulation dynamics to, through a scaling-up approach, create a dispersal model (DM) which can be combined with an ENM. Although some software packages are available providing similar functionalities, MetaLandSim offers a good balance between data requirement and output quality. It assumes that the focal species occupies a clearly defined network of habitat patches, in which local extinction and colonization events are common. The final result, after combining the output of MetaLandSim with that of an ENM, is a set of model's projections considering the ecological needs of the species (ENM), and its dispersal ability and landscape configuration (DM). This output allows the identification of areas with adequate climatic conditions to the species (with the ENM), which are accessible (using the DM). We demonstrate this approach with the Cabrera vole (Microtus cabrerae), a globally " near threatened " species endemic to the Iberian Peninsula, Western Europe. It has a metapopulation-like spatial structure, occupying habitat networks relatively easy to identify and map at the landscape scale. Results highlighted that, under two climate change scenarios, the species' range will suffer substantial fragmentation, characterized by a strong reduction of suitable areas, which is strengthened when imposing accessibility restrictions. Overall, this study provides a useful illustrative example of the synergistic effects of climate and landscape change on potential range shifts in metapopulations. We suggest that our simulation modelling approach provides a useful framework improving such predictions for a broad range of species, and developing effective conservation planning for metapopulations under both climate and landscape change.

1Predator conservation management requires detailed understanding of the ecological circumstances associated with predation, especially that on economically valuable prey. We examined the mechanisms behind Bonelli's eagle Hieraaetus fasciatus predation on prey of domestic origin, using dietary data from 22 pairs breeding in south-west Portugal (1992–2001) together with information on landscape composition and prey availability.2Numerically, 42·7% (37·7% in biomass) of eagle prey comprised domestic species, about 70% of which were rural pigeons Columba livia and the remainder were racing pigeons Columba livia and domestic fowl Gallus gallus. Rabbits Oryctolagus cuniculus, red-legged partridges Alectoris rufa and jays Garrulus glandarius were the most frequent wild prey (43·1%; 50·8% in biomass). This dietary pattern was remarkably stable over a decade, but within each year the intake of pigeons almost halved over the course of the breeding season.3Landscape composition significantly affected the dietary proportion of wild and domestic prey items. This was particularly evident in territories dominated by eucalyptus Eucalyptus globulus plantations, where there was reduced consumption of rural pigeons and partridges, an increased intake of minor avian prey items and greater diversity in the diet overall.4Bonelli's eagles showed type II functional responses while preying on the most important wild (rabbit) and domestic (rural pigeon) prey, although the former was much stronger. Eagle predation on rabbits declined with increasing abundance of pigeons, and vice versa, but there was no switching in the traditional sense, as selection between these two species was inversely frequency dependent.5Synthesis and applications. Predation by Bonelli's eagle on domestic pigeons results from a combination of high vulnerability of the pigeons to eagles and a shortage of key wild prey such as rabbits and partridges, especially during the early breeding season. Given the relatively low economic value of rural pigeons and their importance in the diet of Bonelli's eagles, they could probably be used as a conservation tool to enhance food resources in breeding territories and to deflect predation from more valuable prey such as partridges and racing pigeons.Predator conservation management requires detailed understanding of the ecological circumstances associated with predation, especially that on economically valuable prey. We examined the mechanisms behind Bonelli's eagle Hieraaetus fasciatus predation on prey of domestic origin, using dietary data from 22 pairs breeding in south-west Portugal (1992–2001) together with information on landscape composition and prey availability.Numerically, 42·7% (37·7% in biomass) of eagle prey comprised domestic species, about 70% of which were rural pigeons Columba livia and the remainder were racing pigeons Columba livia and domestic fowl Gallus gallus. Rabbits Oryctolagus cuniculus, red-legged partridges Alectoris rufa and jays Garrulus glandarius were the most frequent wild prey (43·1%; 50·8% in biomass). This dietary pattern was remarkably stable over a decade, but within each year the intake of pigeons almost halved over the course of the breeding season.Landscape composition significantly affected the dietary proportion of wild and domestic prey items. This was particularly evident in territories dominated by eucalyptus Eucalyptus globulus plantations, where there was reduced consumption of rural pigeons and partridges, an increased intake of minor avian prey items and greater diversity in the diet overall.Bonelli's eagles showed type II functional responses while preying on the most important wild (rabbit) and domestic (rural pigeon) prey, although the former was much stronger. Eagle predation on rabbits declined with increasing abundance of pigeons, and vice versa, but there was no switching in the traditional sense, as selection between these two species was inversely frequency dependent.Synthesis and applications. Predation by Bonelli's eagle on domestic pigeons results from a combination of high vulnerability of the pigeons to eagles and a shortage of key wild prey such as rabbits and partridges, especially during the early breeding season. Given the relatively low economic value of rural pigeons and their importance in the diet of Bonelli's eagles, they could probably be used as a conservation tool to enhance food resources in breeding territories and to deflect predation from more valuable prey such as partridges and racing pigeons.