Efficient Markov chain Monte Carlo sampling for hierarchical hidden Markov models

Traditional Markov chain Monte Carlo (MCMC) sampling of hidden Markov models (HMMs) involves latent states underlying an imperfect observation process, and generates posterior samples for top-level parameters concurrently with nuisance latent variables. When potentially many HMMs are embedded within a hierarchical model, this can result in prohibitively long MCMC runtimes. We study combinations of existing methods, which are shown to vastly improve computational efficiency for these hierarchical models while maintaining the modeling flexibility provided by embedded HMMs. The methods include discrete filtering of the HMM likelihood to remove latent states, reduced data representations, and a novel procedure for dynamic block sampling of posterior dimensions. The first two methods have been used in isolation in existing application-specific software, but are not generally available for incorporation in arbitrary model structures. Using the NIMBLE package for R, we develop and test combined computational approaches using three examples from ecological capture–recapture, although our methods are generally applicable to any embedded discrete HMMs. These combinations provide several orders of magnitude improvement in MCMC sampling efficiency, defined as the rate of generating effectively independent posterior samples. In addition to being computationally significant for this class of hierarchical models, this result underscores the potential for vast improvements to MCMC sampling efficiency which can result from combinations of known algorithms.     

A new heuristic for capturing the complexity of multimodal signals

Many animal signals are inherently multimodal, engaging more than one of the receiver’s sensory systems simultaneously, and it is the interaction between the two modalities that determines the signal’s function (s) and efficacy. It is hence necessary to quantify the effect of each modality relative to the other in order to fully understand animal communication. We have developed a new heuristic to aid in the identification and interpretation of the many distinct ways in which signals in multiple sensory modalities interact. Our approach represents natural variation in signal production for each modality and uses these to generate three-dimensional receiver response surface plots that map the relationships among the signal components and receiver behavior. We accommodate the extant hypotheses for the interactions between modalities, each of which makes a clear prediction about the shape of the response surface, and extend previous theory by considering new phenomena.     

Male coalitions and female behaviour affect male mating success independent of dominance rank and female receptive synchrony in wild Barbary macaques

Dominant mammalian males should gain a reproductive advantage due to their greater fighting abilities. However, the extent to which they can monopolise access to females varies across species. In primates and recently other mammalian species, the Priority of Access (PoA) model is commonly used to measure the degree to which male rank and female receptive synchrony affect mating skew. Few studies have examined the factors which lead to deviations from the expectations of the model. Here, we investigate male mating skew in wild Barbary macaques (Macaca sylvanus). We examined four of the main factors which affect male mating success: the roles of male rank, female receptive synchrony, coalitionary activity and female behaviour. We found that male mating was skewed up the hierarchy, but there was a large deviation from the PoA model's expectations with high-ranked males not gaining as big a share as expected. Females frequently initiated sexual encounters, predominantly with mid-ranked males, increasing their mating success. Male coalitionary activity independently increased mating success. Frequent associations with females were costly to males as they were the targets of bridging coalitions, decreasing future mating opportunities for the targets. High-ranking males did not increase their mating success directly through bridging coalitions but acted to dilute the effects of female behaviour. By examining different factors affecting mating skew, we are able to show that alternative male and female mating strategies are effective in reducing the monopolisation potential of the dominant male.     

Ten years of land cover change on the California coast detected using landsat satellite image analysis: part 1—Marin and San Francisco counties

Landsat satellite imagery was analyzed to generate a detailed record of 10 years of vegetation disturbance and regrowth for Pacific coastal areas of Marin and San Francisco Counties. The Landsat Ecosystem Disturbance Adaptive Processing System (LEDAPS) methodology, a transformation of Tasseled-Cap data space, was applied to detected changes in perennial coastal shrubland, woodland, and forest cover from 1999 to 2009. Results showed several principal points of interest, within which extensive contiguous areas of similar LEDAPS vegetation change (either disturbed or restored) were detected. Regrowth areas were delineated as burned forest areas in the Point Reyes National Seashore (PRNS) from the 1995 Vision Fire. LEDAPS-detected disturbance patterns on Inverness Ridge, PRNS in areas observed with dieback of tanoak and bay laurel trees was consistent with defoliation by sudden oak death (Phytophthora ramorum). LEDAPS regrowth pixels were detected over much of the predominantly grassland/herbaceous cover of the Olema Valley ranchland near PRNS. Extensive restoration of perennial vegetation cover on Crissy Field, Baker Beach and Lobos Creek dunes in San Francisco was identified. Based on these examples, the LEDAPS methodology will be capable of fulfilling much of the need for continual, low-cost monitoring of emerging changes to coastal ecosystems.     

The invasive gastropod Crepidula fornicata: reproduction and recruitment in the intertidal at its northernmost range in Wales, UK, and implications for its secondary spread

The establishment and spread of a non-native species in an introduced range depends to a large extent on the performance of the species under the prevailing environmental conditions. The spawning, larval and spatfall periods of the invasive gastropod Crepidula fornicata were monitored in the intertidal zone at its northernmost range in Wales, UK, between February 2010 and January 2011. The duration of the reproductive season was similar to that recorded from more southerly European populations. Spawning and larval release occurred throughout most of the year even at low seawater temperatures of <7 °C, but benthic recruitment was observed over a much shorter period at seawater temperatures >16 °C. Recruitment was low and likely controlled by post-settlement mortality. These observations suggest that C. fornicata’s northwards spread in Welsh waters will not be limited by seawater temperature negatively affecting reproduction, but by processes acting after larval release. These data show the importance of incorporating settlement and post-settlement processes into studies on recruitment success when aiming to predict the potential spread of a potentially harmful invader such as C. fornicata.     

Transient dynamics and the destabilizing effects of prey heterogeneity

The presence of prey heterogeneity and weakly interacting prey species is frequently viewed as a stabilizer of predator-prey dynamics, countering the destabilizing effects of enrichment and reducing the amplitude of population cycles. However, prior model explorations have largely focused on long-term, dynamic attractors rather than transient dynamics. Recent theoretical work shows that the presence of prey that are defended from predation can have strongly divergent effects on dynamics depending on time scale: prey heterogeneity can counteract the destabilizing effects of enrichment on predator-prey dynamics at long time scales but strongly destabilize systems during transient phases by creating long periods of low predator/prey abundance and increasing extinction probability (an effect that is amplified with increasing enrichment). We tested these general predictions using a planktonic system composed of a zooplankton predator and multiple algal prey. We first parameterized a model of our system to generate predictions and tested these experimentally. Our results qualitatively supported several model predictions. During transient phases, presence of defended algal prey increased predator extinctions at low and high enrichment levels compared to systems with only a single edible prey. This destabilizing effect was moderated at higher dilution rates, as predicted by our model. When examining dynamics beyond initial oscillations, presence of the defended prey increased predator-prey temporal variability at high nutrient enrichment but had no effect at low nutrient levels. Our results highlight the importance of considering transient dynamics when assessing the role of stabilizing factors on the dynamics of food webs.