Nearby Patches Research Articles

Deer in the agriculture-forest matrix: Interacting effects of land uses on browsing pressure on trees

Cervids inhabit mosaic landscapes that are strongly influenced by human land use and actions. Forestry and agriculture influence both the foodscape and the risk landscape, thus influencing how cervids respond to resource-risk trade-offs. Furthermore, these land use types interact to shape cervid use of the landscape. For example, cervids can forage in open agricultural fields providing high quality forage during the night, while using nearby forest patches to reduce risk during the day. Hence, actions conducted in one of these land use types will likely influence the use of the other by cervids. In this study, we investigate how agricultural and silvicultural practices interact to shape cervid use of the forest and their impact on commercially valuable trees. More specifically, we investigate the effect of growing nutrient dense crops in mixed forest-agriculture landscapes. We hypothesized that the sowing of such crops may not only attract cervids to the agricultural parts of the landscape, but also lead to increased browsing levels in adjacent forest areas where cervids seek cover. Moreover, such high quality crops may not only influence the number of animals, but also increase the intake rate of fibrous tree saplings of individual cervids to nutritionally balance the high input of easily-digestible macronutrients from the crops. We tested these ideas by conducting a large landscape-scale experiment where we manipulated the foodscape by growing two crop types on 28 fields (14 fields with high-quality oat versus 14 fields with lower-quality grass) and measuring cervid use of tree saplings in the forest patches adjacent these fields. We also investigated how forage availability in the forest patches and the distance from field edge affected browsing on tree saplings. Finally, we compared the effects of crop type and forest forage availability on the browsing pressure on coniferous versus deciduous tree species. We studied this in a multispecies ungulate system, where four cervid species occur sympatrically. We show that the type of crop sown influenced browsing pressure on tree saplings in nearby forests, with higher browsing levels in forests surrounding oat compared to grass fields. Moreover, browsing pressure in these forests was influenced by the availability of young trees in the forest stands, where more young trees led to reduced browsing pressure on individual saplings. Our study emphasizes that human practices that shape features of the foodscape in one land use type will influence the impact cervids have in another land use type. Hence, our study highlights the importance of considering the actions and impacts of different land uses simultaneously for predicting cervid foraging patterns in mixed forest-agriculture landscapes.

Scale Optimization Using Evolutionary Reinforcement Learning for Object Detection on Drone Imagery

Object detection in aerial imagery presents a significant challenge due to large scale variations among objects. This paper proposes an evolutionary reinforcement learning agent, integrated within a coarse-to-fine object detection framework, to optimize the scale for more effective detection of objects in such images. Specifically, a set of patches potentially containing objects are first generated. A set of rewards measuring the localization accuracy, the accuracy of predicted labels, and the scale consistency among nearby patches are designed in the agent to guide the scale optimization. The proposed scale-consistency reward ensures similar scales for neighboring objects of the same category. Furthermore, a spatial-semantic attention mechanism is designed to exploit the spatial semantic relations between patches. The agent employs the proximal policy optimization strategy in conjunction with the evolutionary strategy, effectively utilizing both the current patch status and historical experience embedded in the agent. The proposed model is compared with state-of-the-art methods on two benchmark datasets for object detection on drone imagery. It significantly outperforms all the compared methods. Code is available at https://github.com/UNNC-CV/EvOD/.

Efficient and Accurate Image Classification via Spatial Pyramid Matching and SURF Sparse Coding

Sparsely coded signal space representations do well in feature quantization. Instead of using standard vector quantization, the suggested method uses selective sparse coding to assemble the most important features of the appearance descriptors of nearby image patches. Inadequate coding also enables adjacent max pooling on some spatial scales, which, unlike the setup of average pooling in a histogram, links interpretation with scale invariance. The acquired visual illustration is the key contribution of this research; it performs outperform with linear-SVM, improves the model training's, which in turn speeds up testing with improves accuracy. The efficacy of the method we have employed has been substantiated through a series of experiments conducted on diverse datasets. Since top-performing image classification systems heavily rely on nonlinear SPM in mean of vector quantization, the trustworthy recommended linear SPM greatly increases the use of larger sets of training data. The method given herein deduces that the sparse coding of SURF feature’s function hampered a more comprehensive local appearance descriptor for general-purpose image processing. Experiments and comparisons are conducted on standard datasets such as Caltech-101, FTVL, and Corel-1000, using state-of-the-art techniques and descriptors. When compared over several other image categories and descriptors, the method provided here comes out on top.

Dynamics of soil organic carbon pools following conversion of savannah to cocoa agroforestry systems in the Centre region of Cameroon

Afforestation of gramineous-woody savannah with cocoa agroforestry systems (cAFS) is a common farmer practice in Cameroon considered as sustainable. Nevertheless, the effects of afforestation of savannah with cAFS on soil organic carbon (SOC) turnover and content, and the factors controlling SOC accumulation and stabilization are unknown. SOC content at 0–10 cm soil layer, and SOC distribution in soil particle size fractions (0–20 μm fraction considered as mineral-associated organic carbon, MAOC; 50–2000 μm considered as particulate organic carbon, POC; and 20–50 μm), were compared in different systems settled on degraded savannah (orthic ferralsols). These systems included annual cropland (≈ 5 years old), cocoa monoculture (≈10 years old), and cAFS (from 20 to 60 years old) including different shade tree species such as Albizia adianthifolia, Canarium schweinfurthii, Dacryodes edulis, Milicia excelsa and Ceiba pentandra. Savannah and nearby secondary forest patches were also included in the design as controls. Soil 13C was analysed to investigate the soil carbon turnover after afforestation (C3 plants) of gramineous savannah (C4 plants).SOC significantly increased in the 0–10 cm depth from 10.6 ± 3.1 g C kg−1 in degraded savannah to 17.9 ± 5.6 g C kg−1 in cAFS reaching similar levels as in nearby secondary forests (16.3 ± 5.8 g C kg−1), while annual cropland and cocoa monoculture presented a non-significant decrease in SOC content. These changes were due to rapid loss of SOC derived from savannah plants (C4) – about 76% within the first 15 years after conversion, and higher gain of SOC derived from C3 plants in cAFS than in the other land uses (e.g. from 3.4 ± 1.5 g C kg−1 in savannah to 17.8 ± 5.7 g C kg−1 in cAFS). This SOC enrichment in cAFS was distributed in POC (64%), MAOC (30%) and the intermediate 20–50 μm soil fraction (6%). The higher annual litter input accumulated on a longer period in cAFS (20 to 60 years) than in cocoa monoculture (10 years) concomitant with the lower litter recalcitrance of associated trees compared to cocoa could explain the higher enrichment of SOC in all fractions in cAFS. The soil pH and exch. Ca2+ differed under the different shade tree species, and were positively correlated to SOC content. The highest contents of soil exch. Ca2+ induced by Ceiba and Milicia in the top 10 cm soil layer could contribute to increase SOC enrichment under those species through soil aggregation and related C stabilization. We found no strong evidence of the effect of soil texture on additional soil carbon accumulation in cAFS, especially for the more stable C pool (MAOC). Our results evidenced that savannah afforestation with cAFS appears as a valuable option for top soil carbon enrichment and should consider tree species associated to cocoa to enhance soil C sequestration, soil quality and cocoa production sustainability.

Food availability and population parameters for squirrels differ even in neighbouring urban parks

Some species occur in higher densities in cities than in natural habitats, despite the variety of urban stressors. Urban fragments can be extremely heterogeneous though, and species’ responses might vary between urban patches. However, only few studies examine mammalian populations from different fragments in the core of cities, where stressors are at their maximum. The Eurasian red squirrel (Sciurus vulgaris) is highly abundant in urban green spaces and readily makes use of supplemental food sources. We examined natural and supplemental food sources as well as squirrel populations in three small parks in an urban core. We found pronounced differences in natural and supplemental food availability between the parks. Squirrel population densities were generally higher at all three urban sites when compared to a geographically close located natural forest, but with variation between the parks. The squirrel density in the smallest park in the summer season was 42 times higher (index of minimum number alive: 8.8 individuals/ha) than in the forest, whereas it was only 20 or respectively 16 times higher in the two other parks. Individuals from the park with the highest population density had access to the highest amount of supplemental food sources. However, side effects from non-natural food items, e.g., related to nutrition, also need to be taken into consideration and might counteract the advantages of supplemental food sources. This study shows that urban effects can vary even at the smallest scale between nearby urban habitat patches and this needs to be considered when drawing conclusions about the effects of urbanisation and deciding about conservation actions.

Warmer springs increase potential for temporal reproductive isolation among habitat patches in subalpine flowering plants

Abstract Flowering phenology can vary considerably even at fine spatial scales, potentially leading to temporal reproductive isolation among habitat patches. Climate change could alter flowering synchrony, and hence temporal isolation, if plants in different microhabitats vary in their phenological response to climate change. Despite the importance of temporal isolation in determining patterns of gene flow, and hence population genetic structure and local adaptation, little is known about how changes in climate affect temporal isolation within populations. Here, we use flowering phenology and floral abundance data of 50 subalpine plant species over 44 years to test whether temporal isolation between habitat patches is affected by spring temperature. For each species and year, we analysed temporal separation in peak flowering and flowering overlap between habitat patches separated by 5–950 m. Across our study species, warmer springs were associated with more temporal differentiation in flowering peaks among habitat patches, and less flowering overlap, increasing potential for temporal isolation within populations. Synthesis. By reducing opportunities for mating among plants in nearby habitat patches, our results suggest that warmer springs may reduce opportunities for gene flow within populations, and, consequently, the capacity of plant populations to adapt to environmental changes.

The Influence of Invasive Alien Plants on Umbrella Butterflies of the GenusPhengarisand Diversity of Heteroptera True Bugs

AbstractThe Natura 2000 initiative and National Parks are the most important forms of nature protection in European countries. However, conservation efforts are often hampered by biological invasions, which are some of the most serious environmental problems. The aim of this study was to assess, in protected areas, the impacts of invasive plants – alienSolidagospp. onPhengarisbutterflies andHeracleum mantegazzianumon Heteroptera true bugs.Ph. teleiusandPh. nausithouswere surveyed in the Jaworzno Meadows Natura 2000 site using the Capture-Mark-Recapture method. Heteroptera were surveyed using transect captures in Magurski National Park. The sizes of local populations of the butterflies were lower in patches with higher goldenrod cover. A high cover of the host plant may mitigate the negative impact of goldenrod during the early stage of invasion. In the case of poor habitat quality, nearby unprotected habitat patches act as ecological traps for butterflies due to the application of mowing during the butterfly flight period. The invasive giant hogweed had the most significant impact on reducing the number of Heteroptera individuals. Its influence was also observed at the assemblage level. The negative impact of hogweed has been detected despite the application of the first eradication treatments of this plant. Success in eradicating hogweed may be limited due to the spread of the plants from locations where removal is not applied.

Recolonization of secondary forests by locally extinct fauna through the lens of range expansion: Four open questions

Recolonization of secondary forests happens when individuals disperse from a nearby source old-growth forest populations. This pattern of recolonization could be (a) the result of a random subset of individuals dispersing and colonizing nearby secondary habitats. Instead, the set of recolonizing individuals may not be random but have a particular set of characteristics. (b) Old-growth source populations could show spatial sorting where highly dispersive individuals (those with larger limbs, or exploratory and aggressive behavior) are overrepresented in the forest patch edges and more likely to colonize nearby patches. These are often known as “pull” expansions because highly dispersive individuals living at the edge of the source population are the ones “pulling” the expansion. Alternatively, (c) because old-growth populations are expected to be at carrying capacity recolonization may be driven by subordinate individuals that cannot outcompete dominant conspecifics and disperse looking for alternative territories. This is the case of “push” expansions when dispersal is driven by these subordinate individuals that are pushed away due to density dependence. Data sharing is not applicable to this article as no data were created or analyzed in this study.

Fire effects on soil seed banks under different woody plant species in Mazandaran province, Iran

Soil seed banks play a major role in the post-fire regeneration of semi-arid mountain grasslands. Plant species present before fire can determine the soil seed bank (SSB) characteristics in fire-prone ecosystems. However, it remains unclear how the density and species composition of the SSB under the canopy cover of specific shrub species may be affected by fire. Therefore, we designed a study to test the effects of prescribed burning on the viable SSBs under the canopies of two shrub species: Berberis integerrima (with open canopies) and Onobrychis cornuta (with dense canopies). We selected 20 study sites that included separate patches of the two shrub species and nearby herbaceous patches as control. Soil sampling was carried out pre- and post-burning of shrub canopies and the control patch. Soil samples were collected at two depths: 0–5 cm and 5–10 cm, and transported to the greenhouse for germination experiments. The results showed that canopy fire of shrubs decreased SSB density and species richness. However, the effects of shrub burning on SSB were species-specific and these reductions were more pronounced for canopy burning of O. cornuta than for B. integerrima. Total SSB densities decreased by 61% for B. integerrima and 71% for O. cornuta after canopy fire at soil depth of 0–5 cm. At soil depth of 5–10 cm, total SSB density decreased by 45% under O. cornuta after canopy fire, while canopy fire of B. integerrima did not affect SSB density. Herbaceous patch burning had no statistically significant effects on SSB density at soil depths of 0–5 cm and 5–10 cm. The comparison of quantitative and qualitative similarity indices between pre- and post-fire species composition of shrub canopy showed that the statistically significant effect of B. integerrima burning on SSB composition was less pronounced compared with O. cornuta. This study indicated that seeds stored within the soil under certain shrub species are at a high risk of mortality during fire. Thus, successful post-fire recovery does not rely on the seed bank reservoir under these shrubs.

Habitat-Patch Occupancy of Western Screech-Owls in Suburban Southern California

Many predators can adapt to human-altered habitats, but varying responses to human activity mean that the findings of studies of a species in one region may not apply even to related species at other locations. The Eastern Screech-Owl (Megascops asio) has been extensively studied in suburban environments, but comparatively little research has investigated to what extent the closely related Western Screech-Owl (M. kennicottii) persists in developed areas. I studied Western Screech-Owl occupancy in the town of Alpine in the foothills of San Diego County, California, finding the species at the majority of sites sampled, more likely in larger wooded patches and in those adjacent to undeveloped chaparral. The distance to and size of nearby wooded patches did not statistically influence occupancy. This study confirmed that the Western Screech-Owl can persist in suburban residential areas as does its eastern relative. Ensuring that patches of woodland remain larger than the minimum modeled threshold for the Western Screech-Owl’s occupancy, 4.2 ha, will be important for its persistence in the suburban sprawl of Alpine.

Arbuscular mycorrhizal root colonization depends on the spatial distribution of the host plants

Despite their ubiquity in terrestrial ecosystems, arbuscular mycorrhizal fungi (AMF) experience dispersion constraints and thus depend on the spatial distribution of the plant hosts. Our understanding of fungal-plant interactions with respect to their spatial distributions and implications for the functioning of the symbiosis remain limited. We here manipulated the location of habitat patches of Medicago lupulina in two experiments to explore the responses of AMF root colonization and extraradical hyphae. We tested the specific hypothesis that AMF-plant habitats high in connectance would stimulate root colonization and induce denser functional root colonization (colonization rate of arbuscules plus coils) because of higher propagule availability between nearby host plant patches (experiment 1). In experiment 2, we anticipated similar responses in mixed habitats of different soil fertility, namely phosphorus-fertilized or unfertilized soil, and anticipated a higher density of extraradical hyphae in the soil connecting the habitats with increased functional root colonization. In agreement with our hypothesis, we found the highest total and functional root colonization in unfragmented micro-landscapes, describing landscapes that occur within a spatial scale of a few centimeters with the AMF-plant habitats positioned adjacent to each other. In the second experiment, overdispersed micro-landscapes promoted functional root colonization. This study provides experimental evidence that the spatial distribution of habitats can determine AMF abundance at the microscale.

Active restoration in South African fynbos – A long-term perspective from the Agulhas Plain

In 2019 we re-surveyed a site on the Agulhas Plain, South Africa, which had been part of a restoration project conducted in 2008. The site was invaded by invasive trees, mainly Australian Acacia species. In the original restoration program, the site had been burned and various treatments (i.e. sowing of competitive, fast-growing native species) were conducted to determine the effectiveness of different restoration strategies. The sowing of native species to assist the recovery process is defined as an active restoration method, whereas passive restoration only removes stressors for the ecosystem (e.g. removal of the invader) and leaves it to recover by itself. For this study, we conducted a vegetation survey at the previously restored sites and compared them with a nearby reference site where no intervention had taken place and which is characterised by near-pristine native fynbos vegetation. Secondly, we compared the new data from 2019 with older data from 2010. However, a decade after the initial restoration effort, both active and passively treated plots had not achieved the same condition (species richness, diversity, evenness) as the reference site. The results also show that active restoration is most likely unnecessary in fynbos restoration projects where there are nearby patches with native vegetation or the existing native seed bank is of sufficient size. It seems that it is more important to prevent the reinvasion of invasive alien plants so that the vegetation can recover over time without further disruption.

On the reproductive biology of the critically threatened Alagoas Antwren (Myrmotherula snowi)

The Alagoas Antwren (Myrmotherula snowi) is a Critically Endangered endemic bird restricted to the Pernambuco Center of Endemism. It is currently restricted to a single forest fragment—the Murici Ecological Station in Alagoas. Its population is currently estimated at less than 30 individuals. Moreover, little is known about the natural history of this species. Since 2010 this bird has been annually monitored, but its nests have never been found. We carried out 131 expeditions to this forest fragment between August 2010 and November 2018. During this period, we captured and marked 11 adults and 7 juveniles. However, it was not until 9 November 2018 that the first nest was discovered. The nest was a deep cup composed of black fungal rhizomorphs with small fragments of dried leaves on the outer side. It was suspended on a horizontal fork of a young plant, 1.7 m above the ground. The eggs were oval in shape, light beige in color with reddish patches distributed throughout the egg and most concentrated in the rhombic pole. The eggs measured 1.53 × 1.02 and 1.54 × 1.03 cm. The nest was monitored, but on 25 November 2018 we found the nest empty; we deduced that it had been preyed upon. Subsequently, we found 3 more nests. However, in our efforts over the most recent 11 expeditions (1 year and 3 months after the original nest discovery until February 2020), we failed to find any additional nests. Our records of juvenile individuals indicate they are still reproducing in the area and represent a possibility to recover this species. However, the strong predation pressure in the area and the absence of mixed flocks are concerning. We urge actions to protect its nests, recompose mixed flocks, translocate individuals to nearby habitat patches, and attempt captive breeding, among efforts that should be evaluated.

Pearls on a String: Numerous Stellar Clusters Strung Along the Same Orbit

Abstract Stars originate from the dense interstellar medium, which exhibits filamentary structure to scales of ∼1 kpc in galaxies like our Milky Way. We explore quantitatively how much resulting large-scale correlation there is among different stellar clusters and associations in orbit-phase space, characterized here by actions and angles. As a starting point, we identified 55 prominent stellar overdensities in the 6D space of orbit (actions) and orbit-phase (angles) among the ∼2.8 million stars with radial velocities from Gaia EDR3 and with d ≤ 800 pc. We then explored the orbit-phase distribution of all sample stars in the same orbit patch as any one of these 55 overdensities. We find that very commonly numerous other distinct orbit-phase overdensities exist along these same orbits, like pearls on a string. These “pearls” range from known stellar clusters to loose, unrecognized associations. Among orbit patches defined by one initial orbit-phase overdensity, 50% contain at least 8 additional orbit-phase pearls of 10 cataloged members; 20% of them contain 20 additional pearls. This is in contrast to matching orbit patches sampled from a smooth mock catalog, or offset nearby orbit patches, where there are only 2 (or 5, respectively) comparable pearls. Our findings quantify for the first time how common it is for star clusters and associations to form at distinct orbit-phases of nearly the same orbit. This may eventually offer a new way to probe the 6D orbit structure of the filamentary interstellar medium.

Optimal distributions of central-place foragers: honey bee foraging in a mass flowering crop

AbstractThe ideal-free distribution and central-place foraging are important ecological models that can explain the distribution of foraging organisms in their environment. However, this model ignores distance-based foraging costs from a central place (hive, nest), whereas central-place foraging ignores competition. Different foraging currencies and cooperation between foragers also create different optimal distributions of foragers, but are limited to a simple two-patch model. We present a hybrid model of the ideal-free distribution that uses realistic competitive effects although accounting for distance-based foraging, and test it using honey bees (Apis mellifera L.) foraging in canola fields (Brassica napus L.). Our simulations show that foragers maximizing efficiency (energy profits ÷losses) prioritize distance to their aggregation more than those maximizing net-rate (energy profits ÷time), and that social foragers move to more distant patches to maximize group benefits, meaning that social foragers do not approach an ideal-free distribution. Simulated efficiency-maximizers had a hump-shaped relationship of trip times with distance, spending shorter amounts of time in both nearby and far-away patches. Canola fields were far more attractive to simulated foragers than semi-natural areas, suggesting limited foraging on semi-natural lands during the bloom period of canola. Finally, we found that the observed distribution of honey bees in canola fields most closely resembled the optimal distribution of solitary efficiency-maximizers. Our model has both theoretical and practical uses, as it allows us to model central-place forager distributions in complex landscapes as well as providing information on appropriate hive stocking rates for agricultural pollination.

Coarse-to-Fine Contrastive Self-Supervised Feature Learning for Land-Cover Classification in SAR Images With Limited Labeled Data.

Contrastive self-supervised learning (CSSL) has achieved promising results in extracting visual features from unlabeled data. Most of the current CSSL methods are used to learn global image features with low-resolution that are not suitable or efficient for pixel-level tasks. In this paper, we propose a coarse-to-fine CSSL framework based on a novel contrasting strategy to address this problem. It consists of two stages, one for encoder pre-training to learn global features and the other for decoder pre-training to derive local features. Firstly, the novel contrasting strategy takes advantage of the spatial structure and semantic meaning of different regions and provides more cues to learn than that relying only on data augmentation. Specifically, a positive pair is built from two nearby patches sampled along the direction of the texture if they fall into the same cluster. A negative pair is generated from different clusters. When the novel contrasting strategy is applied to the coarse-to-fine CSSL framework, global and local features are learned successively by forcing the positive pair close to each other and the negative pair apart in an embedding space. Secondly, a discriminant constraint is incorporated into the per-pixel classification model to maximize the inter-class distance. It makes the classification model more competent at distinguishing between different categories that have similar appearance. Finally, the proposed method is validated on four SAR images for land-cover classification with limited labeled data and substantially improves the experimental results. The effectiveness of the proposed method is demonstrated in pixel-level tasks after comparison with the state-of-the-art methods.

Habitat isolation and the cues of three remote predators differentially modulate prey colonization dynamics in pond landscapes.

Recent evidence suggests predators may change colonization rates of prey in nearby predator-free patches as an example of context-dependent habitat selection. Such remote predator effects can be positive when colonizers are redirected to nearby patches (habitat compression), or negative when nearby patches are avoided (risk contagion). However, it is unknown to what extent such responses are predator- and prey-specific and change with increasing distance from predator patches. We evaluated how cues of fish, backswimmers and dragonfly larvae affect habitat selection in replicated pond landscapes with predator-free patches located at increasing distances from a predator patch. We found evidence for risk contagion and compression, but spatial colonization patterns were both predator- and prey-specific. The mosquito Culex pipiens and water beetle Hydraena testacea avoided patches next to patches with dragonfly larvae (i.e. risk contagion). Predator-free patches next to patches with backswimmers were avoided only by mosquitoes. Mosquitoes preferentially colonized patches at some distance from a fish or backswimmer patch (i.e. habitat compression). Colonization patterns of beetles also suggested habitat compression, although reward contagion could not be fully excluded as an alternative explanation. Water beetles preferred the most isolated patches regardless of whether predators were present in the landscape, showing that patch position in a landscape alone affects colonization. We conclude that habitat selection can be a complex product of patch isolation and the combined effects of different local and remote cues complicate current attempts to predict the distribution of mobile organisms in landscapes.

Accelerating Ecological Sciences from Above: Spatial Contrastive Learning for Remote Sensing

The rise of neural networks has opened the door for automatic analysis of remote sensing data. A challenge to using this machinery for computational sustainability is the necessity of massive labeled data sets, which can be cost-prohibitive for many non-profit organizations. The primary motivation for this work is one such problem; the efficient management of invasive species -- invading flora and fauna that are estimated to cause damages in the billions of dollars annually. As an ongoing collaboration with the New York Natural Heritage Program, we consider the use of unsupervised deep learning techniques for dimensionality reduction of remote sensing images, which can reduce sample complexity for downstream tasks and decreases the need for large labeled data sets. We consider spatially augmenting contrastive learning by training neural networks to correctly classify two nearby patches of a landscape as such. We demonstrate that this approach improves upon previous methods and naive classification for a large-scale data set of remote sensing images derived from invasive species observations obtained over 30 years. Additionally, we simulate deployment in the field via active learning and evaluate this method on another important challenge in computational sustainability -- landcover classification -- and again find that it outperforms previous baselines.

Patch selection by bumble bees navigating discontinuous landscapes

Pollen and nectar resources are unevenly distributed over space and bees must make routing decisions when navigating patchy resources. Determining the patch selection process used by bees is crucial to understanding bee foraging over discontinuous landscapes. To elucidate this process, we developed four distinct probability models of bee movement where the size and the distance to the patch determined the attractiveness of a patch. A field experiment with a center patch and four peripheral patches of two distinct sizes and distances from the center was set up in two configurations. Empirical transition probabilities from the center to each peripheral patch were obtained at two sites and two years. The best model was identified by comparing observed and predicted transition probabilities, where predicted values were obtained by incorporating the spatial dimensions of the field experiment into each model’s mathematical expression. Bumble bees used both patch size and isolation distance when selecting a patch and could assess the total amount of resources available in a patch. Bumble bees prefer large, nearby patches. This information will facilitate the development of a predictive framework to the study of bee movement and of models that predict the movement of genetically engineered pollen in bee-pollinated crops.

Shade‐growing practices lessen the impact of coffee plantations on multiple dimensions of ant diversity

Abstract Land use management influence changes in biodiversity beyond the targeted species. Management practices in coffee plantations have shifted from coffee growing below accompanying (shade) trees, to intensified monocultures in which coffee grows fully exposed to the sun. Anthropogenic disturbance causes changes in species composition relative to adjacent natural patches and reduces their biotic heterogeneity. Here, we assessed the impact of coffee plantation management practices on the taxonomical, phylogenetic and functional composition of ant communities, an ecologically dominant group and crucial biological pest controller in these agroecosystems. We hypothesized that shade‐grown coffee plantations would harbour ant communities similar to those of nearby forest patches, but dissimilar to those of intensified monocultures. We surveyed ant diversity in eight shade‐grown coffee farms, eight intensive coffee monocultures and eight forest patches. We used a combination of active and passive sampling methods over two field campaigns spanning 6 months. Our results support our hypothesis for all diversity dimensions. Additionally, ant communities in intensified monocultures were taxonomically and functionally, but not phylogenetically, more homogeneous than those found in forest patches and shade‐grown plantations. Synthesis and applications. Our findings support the idea that practices in shade‐grown coffee plantations buffer the impoverishment of multiple diversity dimensions after forest conversion. Additionally, we identified that leaf‐litter depth and number of twigs mitigate ant diversity loss which, in turn, can favour the presence of potential biocontrol agents. By assessing and integrating multiple biodiversity dimensions into management strategies, farmers and interested parties can minimize future biodiversity and ecosystem service loss.