Bioacoustic classification of avian calls from raw sound waveforms with an open-source deep learning architecture.

The use of autonomous recordings of animal sounds to detect species is a popular conservation tool, constantly improving in fidelity as audio hardware and software evolves. Current classification algorithms utilise sound features extracted from the recording rather than the sound itself, with varying degrees of success. Neural networks that learn directly from the raw sound waveforms have been implemented in human speech recognition but the requirements of detailed labelled data have limited their use in bioacoustics. Here we test SincNet, an efficient neural network architecture that learns from the raw waveform using sinc-based filters. Results using an off-the-shelf implementation of SincNet on a publicly available bird sound dataset (NIPS4Bplus) show that the neural network rapidly converged reaching accuracies of over 65% with limited data. Their performance is comparable with traditional methods after hyperparameter tuning but they are more efficient. Learning directly from the raw waveform allows the algorithm to select automatically those elements of the sound that are best suited for the task, bypassing the onerous task of selecting feature extraction techniques and reducing possible biases. We use publicly released code and datasets to encourage others to replicate our results and to apply SincNet to their own datasets; and we review possible enhancements in the hope that algorithms that learn from the raw waveform will become useful bioacoustic tools.

Mortality predispositions of conifers across western USA.

Conifer mortality rates are increasing in western North America, but the physiological mechanisms underlying this trend are not well understood. We examined tree-ring-based radial growth along with stable carbon (C) and oxygen (O) isotope composition (δ13 C and δ18 O, respectively) of dying and surviving conifers at eight old-growth forest sites across a strong moisture gradient in the western USA to retrospectively investigate mortality predispositions. Compared with surviving trees, lower growth of dying trees was detected at least one decade before mortality at seven of the eight sites. Intrinsic water-use efficiency increased over time in both dying and surviving trees, with a weaker increase in dying trees at five of the eight sites. C starvation was a strong correlate of conifer mortality based on a conceptual model incorporating growth, δ13 C, and δ18 O. However, this approach does not capture processes that occur in the final months of survival. Ultimately, C starvation may lead to increased mortality vulnerability, but hydraulic failure or biotic attack may dominate the process during the end stages of mortality in these conifers.

Giant cacti: isotopic recorders of climate variation in warm deserts of the Americas.

The plant family Cactaceae is considered among the most threatened groups of organisms on the planet. The threatened status of the cacti family has created a renewed interest in the highly evolved physiological and morphological traits that underpin their persistence in some of the harshest subtropical environments in the Americas. Among the most important anatomical features of cacti is the modification of leaves into spines, and previous work has shown that the stable isotope chemistry of cacti spines records potential variations in stem water balance, stress, and Crassulacean acid metabolism (CAM). We review the opportunities, challenges, and pitfalls in measuring δ 13C, δ 2H, and δ 18O ratios captured in spine tissues that potentially reflect temporal and spatial patterns of stomatal conductance, internal to atmospheric CO2 partial pressures, and subsequent patterns of photosynthetic gas exchange. We then evaluate the challenges in stable isotope analysis in spine tissues related to variation in CAM expression, stem water compartmentalization, and spine whole-tissue composition among other factors. Finally, we describe how the analysis of all three isotopes can be used in combination to provide potentially robust analysis of photosynthetic function in cacti, and other succulent-stemmed taxa across broad spatio-temporal environmental gradients.

Progress in Australian dendroclimatology: Identifying growth limiting factors in four climate zones.

Dendroclimatology can be used to better understand past climate in regions such as Australia where instrumental and historical climate records are sparse and rarely extend beyond 100years. Here we review 36 Australian dendroclimatic studies which cover the four major climate zones of Australia; temperate, arid, subtropical and tropical. We show that all of these zones contain tree and shrub species which have the potential to provide high quality records of past climate. Despite this potential only four dendroclimatic reconstructions have been published for Australia, one from each of the climate zones: A 3592year temperature record for the SE-temperate zone, a 350year rainfall record for the Western arid zone, a 140year rainfall record for the northern tropics and a 146year rainfall record for SE-subtropics. We report on the spatial distribution of tree-ring studies, the environmental variables identified as limiting tree growth in each study, and identify the key challenges in using tree-ring records for climate reconstruction in Australia. We show that many Australian species have yet to be tested for dendroclimatological potential, and that the application of newer techniques including isotopic analysis, carbon dating, wood density measurements, and anatomical analysis, combined with traditional ring-width measurements should enable more species in each of the climate zones to be used, and long-term climate records to be developed across the entire continent.

Stable isotopes in leaf water of terrestrial plants.

Leaf water contains naturally occurring stable isotopes of oxygen and hydrogen in abundances that vary spatially and temporally. When sufficiently understood, these can be harnessed for a wide range of applications. Here, we review the current state of knowledge of stable isotope enrichment of leaf water, and its relevance for isotopic signals incorporated into plant organic matter and atmospheric gases. Models describing evaporative enrichment of leaf water have become increasingly complex over time, reflecting enhanced spatial and temporal resolution. We recommend that practitioners choose a model with a level of complexity suited to their application, and provide guidance. At the same time, there exists some lingering uncertainty about the biophysical processes relevant to patterns of isotopic enrichment in leaf water. An important goal for future research is to link observed variations in isotopic composition to specific anatomical and physiological features of leaves that reflect differences in hydraulic design. New measurement techniques are developing rapidly, enabling determinations of both transpired and leaf water δ(18) O and δ(2) H to be made more easily and at higher temporal resolution than previously possible. We expect these technological advances to spur new developments in our understanding of patterns of stable isotope fractionation in leaf water.

The effects of α-cellulose extraction and blue-stain fungus on retrospective studies of carbon and oxygen isotope variation in live and dead trees.

Tree-ring carbon and oxygen isotope ratios from live and recently dead trees may reveal important mechanisms of tree mortality. However, wood decay in dead trees may alter the δ(13)C and δ(18)O values of whole wood obscuring the isotopic signal associated with factors leading up to and including physiological death. We examined whole sapwood and α-cellulose from live and dead specimens of ponderosa pine (Pinus ponderosa), one-seed juniper (Juniperous monosperma), piñon pine (Pinus edulis) and white fir (Abies concolor), including those with fungal growth and beetle frass in the wood, to determine if α-cellulose extraction is necessary for the accurate interpretation of isotopic compositions in the dead trees. We found that the offset between the δ(13)C or δ(18)O values of α-cellulose and whole wood was the same for both live and dead trees across a large range of inter-annual and regional climate differences. The method of α-cellulose extraction, whether Leavitt-Danzer or Standard Brendel modified for small samples, imparts significant differences in the δ(13)C (up to 0.4‰) and δ(18) O (up to 1.2‰) of α-cellulose, as reported by other studies. There was no effect of beetle frass or blue-stain fungus (Ophiostoma) on the δ(13)C and δ(18)O of whole wood or α-cellulose. The relationships between whole wood and α-cellulose δ(13)C for ponderosa, piñon and juniper yielded slopes of ~1, while the relationship between δ(18)O of whole wood and α-cellulose was less clear. We conclude that there are few analytical or sampling obstacles to retrospective studies of isotopic patterns of tree mortality in forests of the western United States.

Seasonal photosynthetic gas exchange and water-use efficiency in a constitutive CAM plant, the giant saguaro cactus (Carnegiea gigantea).

Crassulacean acid metabolism (CAM) and the capacity to store large quantities of water are thought to confer high water use efficiency (WUE) and survival of succulent plants in warm desert environments. Yet the highly variable precipitation, temperature and humidity conditions in these environments likely have unique impacts on underlying processes regulating photosynthetic gas exchange and WUE, limiting our ability to predict growth and survival responses of desert CAM plants to climate change. We monitored net CO(2) assimilation (A(net)), stomatal conductance (g(s)), and transpiration (E) rates periodically over 2 years in a natural population of the giant columnar cactus Carnegiea gigantea (saguaro) near Tucson, Arizona USA to investigate environmental and physiological controls over carbon gain and water loss in this ecologically important plant. We hypothesized that seasonal changes in daily integrated water use efficiency (WUE(day)) in this constitutive CAM species would be driven largely by stomatal regulation of nighttime transpiration and CO(2) uptake responding to shifts in nighttime air temperature and humidity. The lowest WUE(day) occurred during time periods with extreme high and low air vapor pressure deficit (D(a)). The diurnal with the highest D(a) had low WUE(day) due to minimal net carbon gain across the 24 h period. Low WUE(day) was also observed under conditions of low D(a); however, it was due to significant transpiration losses. Gas exchange measurements on potted saguaro plants exposed to experimental changes in D(a) confirmed the relationship between D(a) and g(s). Our results suggest that climatic changes involving shifts in air temperature and humidity will have large impacts on the water and carbon economy of the giant saguaro and potentially other succulent CAM plants of warm desert environments.

Past climate changes and ecophysiological responses recorded in the isotope ratios of saguaro cactus spines.

The stable isotope composition of spines produced serially from the apex of columnar cacti has the potential to be used as a record of changes in climate and physiology. To investigate this potential, we measured the delta(18)O, delta(13)C and F(14)C values of spines from a long-lived columnar cactus, saguaro (Carnegiea gigantea). To determine plant age, we collected spines at 11 different heights along one rib from the stem apex (3.77 m height) to the base of a naturally occurring saguaro. Fractions of modern carbon (F(14)C) ranged from 0.9679 to 1.5537, which is consistent with ages between 1950 and 2004. We observed a very strong positive correlation (r = 0.997) between the F(14)C age of spines and the age of spines determined from direct and repeated height measurements taken on this individual over the past 37 years. A series of 96 spines collected from this individual had delta(18)O values ranging from 38 per thousand to 50 per thousand [Vienna standard mean ocean water (VSMOW)] and delta(13)C values from -11.5 per thousand to -8.5 per thousand [Vienna Peedee belemnite (VPDB)]. The delta(18)O and delta(13)C values of spines were positively correlated (r = 0.45, P < 0.0001) and showed near-annual oscillations over the approximately 15-year record. This pattern suggests that seasonal periods of reduced evaporative demand or greater precipitation input may correspond to increased daytime CO(2) uptake. The lowest delta(18)O and delta(13)C values of spines observed occurred during the 1983 and 1993 El Niño years, suggesting that the stable isotope composition recorded in spine tissue may serve as a proxy for these climate events. We compared empirical models and data from potted experimental cacti to validate these observations and test our hypotheses. The isotopic records presented here are the first ever reported from a chronosequence of cactus spines and demonstrate that tissues of columnar cacti, and potentially other long-lived succulents, may contain a record of past physiological and climatic variation.

Precipitation pulse use by an invasive woody legume: the role of soil texture and pulse size.

Plant metabolic activity in arid and semi-arid environments is largely tied to episodic precipitation events or "pulses". The ability of plants to take up and utilize rain pulses during the growing season in these water-limited ecosystems is determined in part by pulse timing, intensity and amount, and by hydrological properties of the soil that translate precipitation into plant-available soil moisture. We assessed the sensitivity of an invasive woody plant, velvet mesquite (Prosopis velutina Woot.), to large (35 mm) and small (10 mm) isotopically labeled irrigation pulses on two contrasting soil textures (sandy-loam vs. loamy-clay) in semi-desert grassland in southeastern Arizona, USA. Predawn leaf water potential (psi(pd)), the isotopic abundance of deuterium in stem water (deltaD), the abundance of 13C in soluble leaf sugar (delta13C), and percent volumetric soil water content (theta(v)) were measured prior to irrigation and repeatedly for 2 weeks following irrigation. Plant water potential and the percent of pulse water present in the stem xylem indicated that although mesquite trees on both coarse- and fine-textured soils quickly responded to the large irrigation pulse, the magnitude and duration of this response substantially differed between soil textures. After reaching a maximum 4 days after the irrigation, the fraction of pulse water in stem xylem decreased more rapidly on the loamy-clay soil than the sandy-loam soil. Similarly, on both soil textures mesquite significantly responded to the 10-mm pulse. However, the magnitude of this response was substantially greater for mesquite on the sandy-loam soil compared to loamy-clay soil. The relationship between psi(pd) and delta13C of leaf-soluble carbohydrates over the pulse period did not differ between plants at the two sites, indicating that differences in photosynthetic response of mesquite trees to the moisture pulses was a function of soil water availability within the rooting zone rather than differences in plant biochemical or physiological constraints. Patterns of resource acquisition by mesquite during the dynamic wetting-drying cycle following rainfall pulses is controlled by a complex interaction between pulse size and soil hydraulic properties. A better understanding of how this interaction affects plant water availability and photosynthetic response is needed to predict how grassland structure and function will respond to climate change.

Response of net ecosystem gas exchange to a simulated precipitation pulse in a semi-arid grassland: the role of native versus non-native grasses and soil texture.

Physiological activity and structural dynamics in arid and semi-arid ecosystems are driven by discrete inputs or "pulses" of growing season precipitation. Here we describe the short-term dynamics of ecosystem physiology in experimental stands of native (Heteropogon contortus) and invasive (Eragrostis lehmanniana) grasses to an irrigation pulse across two geomorphic surfaces with distinctly different soils: a Pleistocene-aged surface with high clay content in a strongly horizonated soil, and a Holocene-aged surface with low clay content in homogenously structured soils. We evaluated whole-ecosystem and leaf-level CO2 and H2O exchange, soil CO2 efflux, along with plant and soil water status to understand potential constraints on whole-ecosystem carbon exchange during the initiation of the summer monsoon season. Prior to the irrigation pulse, both invasive and native grasses had less negative pre-dawn water potentials (Psipd), greater leaf photosynthetic rates (Anet) and stomatal conductance (gs), and greater rates of net ecosystem carbon exchange (NEE) on the Pleistocene surface than on the Holocene. Twenty-four hours following the experimental application of a 39 mm irrigation pulse, soil CO2 efflux increased leading to all plots losing CO2 to the atmosphere over the course of a day. Invasive species stands had greater evapotranspiration rates (ET) immediately following the precipitation pulse than did native stands, while maximum instantaneous NEE increased for both species and surfaces at roughly the same rate. The differential ET patterns through time were correlated with an earlier decline in NEE in the invasive species as compared to the native species plots. Plots with invasive species accumulated between 5% and 33% of the carbon that plots with the native species accumulated over the 15-day pulse period. Taken together, these results indicate that system CO2 efflux (both the physical displacement of soil CO2 by water along with plant and microbial respiration) strongly controls whole-ecosystem carbon exchange during precipitation pulses. Since CO2 and H2O loss to the atmosphere was partially driven by species effects on soil microclimate, understanding the mechanistic relationships between the soil characteristics, plant ecophysiological responses, and canopy structural dynamics will be important for understanding the effects of shifting precipitation and vegetation patterns in semi-arid environments.