Ecosphere最新文献

Non-native annual grasses can dramatically alter fire frequency and reduce forage quality and biodiversity in the ecosystems they invade. Effective management techniques are needed to reduce these undesirable invasive species and maintain ecosystem services. Well-timed management strategies, such as grazing, that are applied when invasive grasses are active prior to native plants can control invasive species spread and reduce their impact; however, anticipating the timing of key phenological stages that are susceptible to management over vast landscapes is difficult, as the phenology of these species can vary greatly over time and space. To address this challenge, we created range-wide phenology forecasts for two problematic invasive annual grasses: cheatgrass (Bromus tectorum), and red brome (Bromus rubens). We tested a suite of 18 mechanistic phenology models using observations from monitoring experiments, volunteer science, herbarium records, timelapse camera imagery, and downscaled gridded climate data to identify the models that best predicted the dates of flowering and senescence of the two invasive grass species. We found that the timing of flowering and senescence of cheatgrass and red brome were best predicted by photothermal time models that had been adjusted for topography using gridded continuous heat-insolation load index values. Phenology forecasts based on these models can help managers make decisions about when to schedule management actions such as grazing to reduce undesirable invasive grasses and promote forage production, quality, and biodiversity in grasslands; to predict the timing of greatest fire risk after annual grasses dry out; and to select remote sensing imagery to accurately map invasive grasses across topographic and latitudinal gradients. These phenology models also have the potential to be operationalized for within-season or within-year decision support.

Bolstering the supply of animal-mediated ecosystem services is an emerging priority in human-altered landscapes. Such services are driven not only by environmental factors that shape communities of species that provide the service but also by the ecological context that affects the behavior of these species. In this study, we used a field experiment to investigate an ecosystem service that depends on resource use behavior—the removal of littered food waste by birds and squirrels in urban green spaces. We first explore how landscape-scale urbanization affects the composition of the litter-removing species community. We then examine two facets of waste removal provisioning—the amount of food removed and the speed of removal—and how they vary across ecological contexts represented by green space type (picnic areas, urban parks, and forest preserves), bird and squirrel abundance, number of people, amount of existing litter, and weather conditions. We found that although landscape-scale urbanization affected the composition of species within green spaces, service provisioning was context-dependent. Littered food removal services were provided at higher rates in park and picnic sites than in forest preserves and the abundance of eastern gray squirrels (Sciurus carolinensis) was a main driver of littered food removal services. Where squirrels were abundant, more food was removed, and food removal began and was completed more quickly. When squirrel abundance is accounted for, removal from picnic areas is higher than park sites, indicating context dependence in this service is likely driven by squirrel behavioral responses to ambient food waste levels in these habitats. This study highlights the role of common urban species in providing a valuable ecosystem service and the importance of ecological context in its supply. Efforts to account for animal-mediated ecosystem services in human-altered landscapes should address the potential for services to be driven by a single species and context-dependent factors that influence behavior.

Throughout the Arctic, ice-affiliated marine mammals constitute local subsistence resources but detrimental effects of declines in their sea ice habitats create a need for harvest sustainability assessments in light of climate change. At the same time, empirical data required for thorough population analysis of these species are often sparse at best, as illustrated by the focal species in this study, ringed seals in Svalbard: the last population survey took place two decades ago (2002–2003), demographic data are limited to age, sex, and reproductive status of a small subset of shot individuals, and harvest reporting is patchy and incomplete. Data sparsity is one of the main reasons why potential biological removal (PBR) became a commonly used tool for assessing sustainability of marine mammal harvests. Herein, we calculated PBR for Svalbard ringed seals using both recommended default parameters and population-specific parameters obtained from an integrated population model (IPM). PBR estimates were highly uncertain, suggesting the number of sustainably harvestable individuals could lie anywhere between 0 and 91, with a substantial chance of any harvest being unsustainable under current environmental conditions and trends. Subsequent population viability analyses (PVAs) further confirmed that the current harvest was likely unsustainable, even in a scenario in which sea ice conditions would not deteriorate (and therefore lower pup survival) further. However, uncertainty in population projections was high, and forecasts thus not ideal for formulating management advice. Better forecasts will require more frequent population surveys and obtaining more knowledge regarding the links between vital rates and environmental conditions, both of which may be facilitated by the adoption of novel technology (e.g., drone monitoring, genetic studies). The modeling framework created in this study can be readily updated with new data as they become available, and can serve as a tool for adaptive management of this and other marine mammal populations.

Urbanization is rapidly transforming coastal landscapes around the world, altering the structure and function of marine, intertidal, and terrestrial ecosystems. In tandem, coastal areas are hotspots for human recreation, leading to shifts in wildlife behavior and activity patterns. Together, urban development and recreational use of wildlife habitats can shape wildlife behavior, abundance, and ecosystem dynamics at different spatial and temporal scales. In this study, we explore the impact of urbanization and human and domestic dog activity on the structure of vertebrate scavenging assemblages and the ecosystem functions they provide in sandy beach ecosystems across 40 km of the central California coast, USA. We surveyed vertebrate scavenging assemblages using baited camera traps on 17 beaches spanning a gradient of coastal urbanization. We found that urbanization extent within small spatial scales (i.e., 1 or 3 km radii of each site) and the rate of beach visitation by domestic dogs or humans were the best additive predictors of assemblage structure. We identified pronounced urbanization-associated shifts in the composition of vertebrate scavenger guilds but found that the rate of carrion processing was more strongly influenced by domestic dog habitat use and diel period. Scavenging activity was substantially lower on beaches with more domestic dogs, suggesting that dogs interfere with critical scavenging ecosystem functions on sandy beaches. Our results underscore the pervasive and nuanced effects of urbanization and recreation on the dynamics of land–sea connectivity and suggest a need for comprehensive consideration of cross-ecosystem linkages in ongoing shoreline conservation and development efforts.

Environmental parameters along elevational gradients affect the number of butterflies and the variety of species. However, which variables play significant roles and how they operate can be difficult to untangle. Here, we examine the relationships between observed butterfly richness (overall and subgroups) at different elevation gradients and remotely sensed environmental variables (vegetation productivity, surface temperature, landscape heterogeneity, and moisture stress) using generalized linear models. We surveyed butterflies with a fixed-point count method in 19 elevation bands within 1600–5200 m above sea level in Manang district, trans-Himalayan region, north-central Nepal. The number of butterflies in each elevation band was studied and estimated, then interpolated across the lowest and highest elevation to estimate butterfly species richness. Then, the selection of models was performed on butterfly richness and elevations to test the best model support based on the lowest value of the Akaike information criterion and a multimodel averaging for other environmental variables. Altogether, 94 butterfly species, representing 20 subfamilies and six families, were recorded throughout the study periods. We obtained cubic model support for overall species richness, Papilionidae, and Hesperiidae, quadratic to Nymphalidae and Pieridae, and the linear model to Lycaenidae. In our study, vegetation productivity was found to have a significant positive impact on butterfly communities. Our study further suggests species richness of Papilionidae and Hesperiidae has a strong positive correlation with surface temperature and landscape heterogeneity and negative associations with moisture stress but other subgroups of butterfly communities including overall species richness showed insignificant relationships with these variables. This study provides significant information related to the responses of montane butterflies to environmental variables along elevational gradients from the Himalayas Nepal. However, further detailed studies on the functional behaviors of butterflies potentially offer more insights into their distribution patterns and ecological relationship in the montane environment.

Estimating the size of animal populations plays an important role in evidence-based conservation and management. Some methods for estimating population size rely on animals being individually identifiable. Traditionally, this has been done by marking physically captured animals, but increasingly, animals with distinctive natural markings are surveyed noninvasively using cameras. Animal reidentification from photographs is usually done manually, which is expensive, laborious, and requires considerable skill. An alternative is to develop computer vision methods that can support or replace the manual identification task. We developed an automated approach using deep learning to identify whether a pair of photographs is of the same individual or not. The core of the approach is a similarity learning network that uses paired convolutional neural networks with a triplet loss function to summarize image pairs and decide whether they are from the same individual. Prior to the main matching step, two additional convolutional neural networks perform image segmentation, cropping the animal object within the image, and orientation prediction, deciding which side of the animal was photographed. We applied the approach to four species, with images of the same individual often spanning several years: systematic surveys of bottlenose dolphins (Tursiops truncatus, 2008–2019) and harbor seals (Phoca vitulina, 2015–2019), a citizen science dataset of western leopard toads (Sclerophrys pantherina, unknown dates), and a publicly available repository of humpback whale images (Megaptera novaeangliae, unknown dates). For these species, our best-performing models were able to identify whether a pair of images were from the same individual or different individuals in 95.8%, 94.6%, 88.2%, and 83.8% of the cases, respectively. We found that triplet loss functions outperformed binary cross-entropy loss functions and that data augmentation and additional manual curation of training data provided small but consistent improvements in performance. These results demonstrate the potential of deep learning to replace or, more likely, support and facilitate manual individual identification efforts.

Climate change is motivating a reassessment of how seeds are selected for reforestation, as rapid environmental change can lead to local maladaptation in trees. Genetic association studies and past seed source climate both have the potential to help identify appropriate planting stock, but these techniques have not been compared and tested as part of an operational planting program. In this study, we combined an analysis of single nucleotide polymorphisms (SNPs) associated with environmental gradients in sugar pine (Pinus lambertiana) with an analysis of post-fire seedling survival and growth in a restoration experiment. Our genotype–environment association (GEA) tests of 92 individuals from varying climates within CA revealed 829 SNPs (out of 300,604) with significant association with climate gradients, especially April snowpack. Of these, 323 either had annotations that suggested potential functional importance or were identified by two different methods. We then built Bayesian models of survival and growth for all seedlings in a separate post-fire planting experiment, to test the relative predictive ability of source elevation (a common proxy for source climate) versus the proportion of seedling alleles expected to be locally advantageous based on GEA. Across three sites within the King Fire scar in Eldorado National Forest in 2017, 2018, and 2019, 1774 seedlings were planted. Of these, 206 had enough green needles in 2020 to allow sample collection, and 161 were successfully genotyped. We found that source elevation was generally better at predicting seedling performance than genotype indices, perhaps because of the limited scope of the association analysis. Seed sources from 500 to 1800 ft (152.4–548.6 m) lower in elevation, and one seed zone further south generally performed as well or better than local seed sources. This result, and those of similar previous studies, suggest that “climate matching” using past climate information for existing seed sourcing units is a reasonable starting point for finding seedlings suited to already-altered planting site climate conditions. However, further tests with more extensive genomic and performance data may improve the utility of genotype information for seed selection.

Many methods have been proposed to model the spatial distribution of a species. While some methods have been specifically designed for this purpose, others are well-known statistical tools that can be used in many scientific fields. In this paper, I propose a new ecological niche model, called the Environmental String Model (ESM), that is based on the concept of environmental string, which is defined as being a combination of environmental variables, with as many nodes as environmental variables. There are two types of environmental strings: (1) the abundance-known string and (2) the abundance-unknown string (or target string) for which an estimation of abundance is searched. The novelty of the model is that it assesses the abundance associated with a target string from nearby abundance-known strings, which preserve the local multidimensional relationships with the target string. The model does not provide an abundance estimate in the absence of data from a similar environment and it can therefore deal with truncated spatial distributions or niches. It is tested in the North Atlantic Ocean on two key copepod species, Calanus finmarchicus and Calanus helgolandicus, which have been monitored by the Continuous Plankton Recorder (CPR) survey for decades. I investigate the influence of variables on model performance. I show that the model reconstructs the mean spatial distribution and seasonal fluctuations in both Calanus well. When compared with generalized linear models (GLMs), generalized additive models (GAMs) and generalized regression neural network (GRNN), the ESM gives the best performance. I propose a number of indicators to evaluate the robustness of estimated abundance in space and time and show how the model may be extended to presence/absence or presence-only data. I think that the ESM could be used to fill gaps in any sampling program such as the CPR survey and many satellite databases (e.g., ocean color and photosynthetically active radiation).

Globally, collisions with vehicles result in millions of animal deaths every year, representing a major issue for wildlife conservation and management. Consequently, and importantly, much research has focused on understanding patterns of animal–vehicle collisions with the aim to reduce roadkill of wildlife. However, existing data on animal–vehicle collisions might also represent a novel opportunity to monitor wildlife populations. For this purpose, we compared data of >1.2 million hunter-shot deer and >40,000 deer–vehicle collisions collected over 11 years in Denmark. We show that deer–vehicle collision data can be useful for population monitoring of roe deer (Capreolus capreolus), fallow deer (Dama dama), and red deer (Cervus elaphus). Roe deer was the most numerous species, representing 90% of observations based both on deer–vehicle collisions and on hunting bag statistics. After accounting for factors related to road infrastructure (road length and density, traffic volume), local (municipality) deer–vehicle collisions were highly correlated with hunting bag data for roe and red deer (Pearson's r > 0.7) but not fallow deer, likely due to biases in hunting bags. Moreover, we used deer–vehicle collision data to map spatiotemporal changes in the distribution of fallow and red deer, and demographic changes in all species. Combined, our results suggest that animal–vehicle collision data can be a useful tool to supplement existing methods for monitoring wildlife populations, which will be relevant for the management of these populations. We point to important shortcomings in both animal–vehicle collision and hunting bag data and provide recommendations on how to improve their accuracy in the future, to be applicable for a broader range of species.

The grassland biome is an important sink for atmospheric methane (CH₄), a major greenhouse gas. There is considerable uncertainty in the grassland CH₄ sink capacity due to diverse environmental gradients in which grasslands occur, and many environmental conditions can affect abiotic (e.g., CH₄ diffusivity into soils) and biotic (e.g., methanotrophy) factors that determine spatial and temporal CH₄ dynamics. We investigated the relative importance of a soil's gas diffusivity versus net methanotroph activity in 22 field plots in seven sites distributed across the US Great Plains by making approximately biweekly measures during the growing seasons over 3 years. We quantified net methanotroph activity and diffusivity by using an approach combining a gas tracer, chamber headspace measurements, and a mathematical model. At each plot, we also measured environmental characteristics, including water-filled pore space (WFPS), soil temperature, and inorganic nitrogen contents, and examined the relative importance of these for controlling diffusivity and net methanotroph activity. At most of the plots across the seven sites, CH₄ uptake rates were consistently greatest when WFPS was intermediate at the plot level. Our results show that variation in net methanotroph activity was more important than diffusivity in explaining temporal variations in net CH₄ uptake, but the two factors were equally important for driving spatial variation across the seven sites. WFPS was a significant predictor for diffusivity only in plots with sandy soils. WFPS was the most important control on net methanotroph activity, with net methanotroph activity showing a parabolic response to WFPS (concave down), and the shape of this response differed significantly among sites. Moreover, we found that the WFPS level at peak net methanotroph activity was strongly correlated with the mean annual precipitation of the site. These results suggest that the local precipitation regime determines unique sensitivity of CH₄ uptake rates to soil moisture. Our findings indicate that grassland CH₄ uptake may be predicted using local soil water conditions. More variable soil moisture, potentially induced through predicted future extremes of rainfall and drought, could reduce grassland CH₄ sink capacity in the future.