Automatic detection of fish scale circuli using deep learning.

Teleost fish scales form distinct growth rings deposited in proportion to somatic growth in length, and are routinely used in fish ageing and growth analyses. Extraction of incremental growth data from scales is labour intensive. We present a fully automated method to retrieve this data from fish scale images using Convolutional Neural Networks (CNNs). Our pipeline of two CNNs automatically detects the centre of the scale and individual growth rings (circuli) along multiple radial transect emanating from the centre. The focus detector was trained on 725 scale images and achieved an average precision of 99%; the circuli detector was trained on 40 678 circuli annotations and achieved an average precision of 95.1%. Circuli detections were made with less confidence in the freshwater zone of the scale image where the growth bands are most narrowly spaced. However, the performance of the circuli detector was similar to that of another human labeller, highlighting the inherent ambiguity of the labelling process. The system predicts the location of scale growth rings rapidly and with high accuracy, enabling the calculation of spacings and thereby growth inferences from salmon scales. The success of our method suggests its potential for expansion to other species.

Variation in the post-smolt growth pattern of wild one sea-winter salmon (Salmo salar L.), and its linkage to surface warming in the eastern North Atlantic Ocean.

Variation in circulus spacing on the scales of wild Atlantic salmon is indicative of changes in body length growth rate. We analyzed scale circulus spacing during the post-smolt growth period for adult one sea-winter salmon (n = 1947) returning to Scotland over the period 1993-2011. The growth pattern of the scales was subjectively and visually categorized according to the occurrence and zonal sequence of three intercirculus spacing criteria ("Slow", "Fast" and "Check" zones). We applied hierarchical time-series cluster analysis to the empirical circulus spacing data, followed by post hoc analysis of significant changes in growth patterns within the 20 identified clusters. Temporal changes in growth pattern frequencies showed significant correlation with sea surface temperature anomalies during the early months of the post-smolt growth season and throughout the Norwegian Sea. Since the turn of the millennium, we observed (a) a marked decrease in the occurrence of continuous Fast growth; (b) increased frequencies of fish showing an extended period of initially Slow growth; and (c) the occurrence of obvious growth Checks or hiatuses. These changes in post-smolt growth pattern were manifest also in decreases in the mean body length attained by the ocean midwinter, as sea surface temperatures have risen.

Hierarchical analysis of wild Atlantic salmon (Salmo salar) fecundity in relation to body size and developmental traits.

Using data from wild Atlantic salmon Salmo salar returning to spawn in seven Scottish rivers, we developed a model of fecundity based on individual body size and key developmental traits. We used a novel approach to model selection which maximises predictive accuracy for application to target river stocks to select the best from a suite of Bayesian hierarchical models. This approach aims to ensure the optimal model within the candidate set includes covariates that best predict out-of-sample data to estimate fecundity in areas where no direct observations are available. In addition to body size, the final model included the developmental characteristics of age at smolting and years spent at sea. Using two independent long-term monitoring datasets, the consequences of ignoring these characteristics was revealed by comparing predictions from the best model with models that omitted them.

Precision and accuracy of Dahl-Lea back-calculated smolt lengths from adult scales of Atlantic salmon Salmo salar.

Using tagged and recaptured Atlantic salmon Salmo salar (n = 106) the present analysis shows that the most commonly applied linear back-calculation method for estimating past length, the Dahl-Lea method, resulted in overestimation of the length of large smolts and underestimation of small smolts. A correction equation (y = 0.53x + 6.23) for estimating true smolt length (y) from lengths back-calculated from adult scale measures (x) to account for these systematic discrepancies is proposed.

Pup mortality in a rapidly declining harbour seal (Phoca vitulina) population.

The harbour seal population in Orkney, off the north coast of Scotland, has reduced by 65% between 2001 and 2010. The cause(s) of this decline are unknown but must affect the demographic parameters of the population. Here, satellite telemetry data were used to test the hypothesis that increased pup mortality could be a primary driver of the decline in Orkney. Pup mortality and tag failure parameters were estimated from the duration of operation of satellite tags deployed on harbour seal pups from the Orkney population (n = 24) and from another population on the west coast of Scotland (n = 24) where abundance was stable. Survival probabilities from both populations were best represented by a common gamma distribution and were not different from one another, suggesting that increased pup mortality is unlikely to be the primary agent in the Orkney population decline. The estimated probability of surviving to 6 months was 0.390 (95% CI 0.297 - 0.648) and tag failure was represented by a Gaussian distribution, with estimated mean 270 (95% CI = 198 - 288) and s.d. 21 (95% CI = 1 - 66) days. These results suggest that adult survival is the most likely proximate cause of the decline. They also demonstrate a novel technique for attaining age-specific mortality rates from telemetry data.

Getting long in the tooth: a strong positive correlation between canine size and heterozygosity in Antarctic fur seals Arctocephalus gazella.

Most studies of heterozygosity-fitness correlations (HFCs) in natural populations relate to fitness traits expressed early in life, whereas traits that are often more difficult to measure such as longevity and adult body size remain elusive. Teeth provide a window on an individual's life history, allowing the reliable estimation of both age and body size. Consequently, we collected paired upper canine teeth and tissue samples from 84 adult male Antarctic fur seals Arctocephalus gazella that died of natural causes at Bird Island, South Georgia. Tooth size is a good predictor of skull and body size both within and across taxa, and we similarly find a strong relationship with skull size in our species. In turn, tooth size is itself predicted strongly by genetic heterozygosity estimated using 9 microsatellites. With only 9 loci, the exact mechanisms involved remain unclear, although the observed pattern appears largely attributable to a small subset of loci, suggesting that associative overdominance rather than inbreeding depression provides the proximate mechanism. In addition, locating these markers in the dog genome reveals proximity to genes involved with fat metabolism and growth. Our study illustrates how canine teeth, and potentially other structures such as tympano-periotic bone, waxy inner earplugs, or otoliths, may be used to explore links between genetic variation and important life-history traits in free-ranging vertebrate populations.