Three critical factors affecting automated image species recognition performance for camera traps

Ecological camera traps are increasingly used by wildlife biologists to unobtrusively monitor an ecosystems animal population. However, manual inspection of the images produced is expensive, laborious, and time‐consuming. The success of deep learning systems using camera trap images has been previously explored in preliminary stages. These studies, however, are lacking in their practicality. They are primarily focused on extremely large datasets, often millions of images, and there is little to no focus on performance when tasked with species identification in new locations not seen during training. Our goal was to test the capabilities of deep learning systems trained on camera trap images using modestly sized training data, compare performance when considering unseen background locations, and quantify the gradient of lower bound performance to provide a guideline of data requirements in correspondence to performance expectations. We use a dataset provided by Parks Canada containing 47,279 images collected from 36 unique geographic locations across multiple environments. Images represent 55 animal species and human activity with high‐class imbalance. We trained, tested, and compared the capabilities of six deep learning computer vision networks using transfer learning and image augmentation: DenseNet201, Inception‐ResNet‐V3, InceptionV3, NASNetMobile, MobileNetV2, and Xception. We compare overall performance on “trained” locations where DenseNet201 performed best with 95.6% top‐1 accuracy showing promise for deep learning methods for smaller scale research efforts. Using trained locations, classifications with <500 images had low and highly variable recall of 0.750 ± 0.329, while classifications with over 1,000 images had a high and stable recall of 0.971 ± 0.0137. Models tasked with classifying species from untrained locations were less accurate, with DenseNet201 performing best with 68.7% top‐1 accuracy. Finally, we provide an open repository where ecologists can insert their image data to train and test custom species detection models for their desired ecological domain.     

Migratory New World Blackbirds (Icterids) Are More Neophobic than Closely Related Resident Icterids

Environments undergo short-term and long-term changes due to natural or human-induced events. Animals differ in their ability to cope with such changes which can be related to their ecology. Changes in the environment often elicit avoidance reactions (neophobia) which protect animals from dangerous situations but can also inhibit exploration and familiarization with novel situations and thus, learning about new resources. Studies investigating the relationship between a species’ ecology and its neophobia have so far been restricted to comparing only a few species and mainly in captivity. The current study investigated neophobia reactions to experimentally-induced changes in the natural environment of six closely-related blackbird species (Icteridae), including two species represented by two distinct populations. For analyses, neophobic reactions (difference in number of birds feeding and time spent feeding with and without novel objects) were related to several measures of ecological plasticity and the migratory strategy (resident or migratory) of the population. Phylogenetic relationships were incorporated into the analysis. The degree of neophobia was related to migratory strategy with migrants expressing much higher neophobia (fewer birds feeding and for a shorter time with objects present) than residents. Furthermore, neophobia showed a relationship to diet breadth with fewer individuals of diet generalists than specialists returning when objects were present supporting the dangerous niche hypothesis. Residents may have evolved lower neophobia as costs of missing out on opportunities may be higher for residents than migrants as the former are restricted to a smaller area. Lower neophobia allows them approaching changes in the environment (e.g. novel objects) quickly, thereby securing access to resources. Additionally, residents have a greater familiarity with similar situations in the area than migrants and the latter may, therefore, initially stay behind resident species.     

Book reviews

John B. Boles (ed.), MASTERS AND SLAVES IN THE HOUSE OF THE LORD: RACE AND RELIGION IN THE AMERICAN SOUTH, 1740–1870, Lexington, KY: The University of Kentucky Press, 1988, 257 pp.

David E. Swift, BLACK PROPHETS OF JUSTICE: ACTIVIST CLERGY BEFORE THE CIVIL WAR, Baton Rouge, Louisiana, LA: Louisiana State University Press, 1989, xv + 384 pp., £31.50.

Alexander B. Murphy, THE REGIONAL DYNAMICS OF LANGUAGE DIFFERENTIATION IN BELGIUM: A STUDY IN CULTURAL‐POLITICAL GEOGRAPHY, Chicago, IL: University of Chicago, Geography Research Paper No. 227, 1988, xiii + 249 pp., $12.00.

Pierre L. van den Berghe, STRANGER IN THEIR MIDST, Denver, CO: University Press of Colorado, 1989, xvii + 300 pp., $24.95.     

Next-generation phenotyping: requirements and strategies for enhancing our understanding of genotype–phenotype relationships and its relevance to crop improvement

More accurate and precise phenotyping strategies are necessary to empower high-resolution linkage mapping and genome-wide association studies and for training genomic selection models in plant improvement. Within this framework, the objective of modern phenotyping is to increase the accuracy, precision and throughput of phenotypic estimation at all levels of biological organization while reducing costs and minimizing labor through automation, remote sensing, improved data integration and experimental design. Much like the efforts to optimize genotyping during the 1980s and 1990s, designing effective phenotyping initiatives today requires multi-faceted collaborations between biologists, computer scientists, statisticians and engineers. Robust phenotyping systems are needed to characterize the full suite of genetic factors that contribute to quantitative phenotypic variation across cells, organs and tissues, developmental stages, years, environments, species and research programs. Next-generation phenotyping generates significantly more data than previously and requires novel data management, access and storage systems, increased use of ontologies to facilitate data integration, and new statistical tools for enhancing experimental design and extracting biologically meaningful signal from environmental and experimental noise. To ensure relevance, the implementation of efficient and informative phenotyping experiments also requires familiarity with diverse germplasm resources, population structures, and target populations of environments. Today, phenotyping is quickly emerging as the major operational bottleneck limiting the power of genetic analysis and genomic prediction. The challenge for the next generation of quantitative geneticists and plant breeders is not only to understand the genetic basis of complex trait variation, but also to use that knowledge to efficiently synthesize twenty-first century crop varieties.     

Seasonal dimorphism in the horny bills of sparrows

Bill size is often viewed as a species‐specific adaptation for feeding, but it sometimes varies between sexes, suggesting that sexual selection or intersexual competition may also be important. Hypotheses to explain sexual dimorphism in avian bill size include divergence in feeding niche or thermoregulatory demands, intrasexual selection based on increased competition among males, or female preference. Birds also show seasonal changes in bill size due to shifts in the balance between growth rate and wear, which may be due to diet or endogenous rhythms in growth. Insight into the function of dimorphism can be gained using the novel approach of digital x‐ray imaging of museum skins to examine the degree to which the skeletal core or the rhamphotheca contribute to overall dimorphism. The rhamphotheca is ever‐growing and ever‐wearing, varying in size throughout life; whereas the skeletal core shows determinant growth. Because tidal marsh sparrows are more dimorphic in bill size than related taxa, we selected two marsh taxa to investigate dimorphism and seasonality in the size of the overall bill, the skeletal core, and the rhamphotheca. Bill size varied by sex and season, with males having larger bills than females, and bill size increasing from nonbreeding to breeding season more in males. Skeletal bill size varied with season, but not sex. The rhamphotheca varied primarily with sex; males had a larger rhamphotheca (corrected for skeletal bill size), which showed a greater seasonal increase than females. The rhamphotheca, rather than the skeletal bill, was responsible for sexual dimorphism in overall bill size, which was particularly well developed in the breeding season. The size of the rhamphotheca may be a condition‐based character that is shaped by sexual selection. These results are consistent with the evidence that bill size is influenced by sexual selection as well as trophic ecology.     

Obesity and early mortality in the United States

Objective:

Although obesity is a serious public health problem, there are few reliable measures of its health hazards in the United States. The objective of this study was to estimate how much earlier mortality is likely to occur for Americans who are obese (body mass index [BMI], ≥ 30).

Design and Methods:

Data from the National Health and Nutrition Examination Survey (NHANES) I (1971–1975), NHANES II (1976–1980), and NHANES III (1988–1994) for 37,632 participants who experienced 8,791 deaths during 15 years of follow‐up were prospectively analyzed. The relative risk of death from all causes and its advancement period, adjusted for covariates, were calculated. Stratification was used to investigate the effects of pre‐existing illness, smoking, and older age on the advancement period.

Results:

Compared to the participants of reference weight (BMI, 23 to <25 kg/m2), mortality was likely to occur 9.44 years (95% confidence interval [CI]: 0.72, 18.16) earlier for those who were obese (BMI, ≥ 30). For overweight (25 to <30 kg/m2), grade 1 obesity (BMI, 30 to <35) and grades 2–3 obesity (BMI, ≥ 35.0), the mortality was likely to occur earlier by 4.40 (?3.90, 12.70), 6.69 (?2.06, 15.43), and 14.16 (3.35, 24.97) years, respectively. These estimates apply to healthy nonsmoker young‐ and middle‐aged (21–55 years) adults, who constituted an estimated 32.8% of Americans with age of >21 years between 1988 and 1994. Without stratifying simultaneously for preexisting illness, smoking, and age, values of the advancement period for obesity were markedly smaller than those observed for healthy nonsmoker young and middle‐aged adults.

Conclusions:

For healthy nonsmokers young‐ and middle‐aged adults who constitute about one‐third of American adults, being obese is likely to hasten mortality by 9.44 years.