Comparative analysis of statistical and machine learning approaches for predicting fish length from otoliths

This study aims to determine the biometric comparison of otoliths of Banggai cardinal fish based on gender and to determine the comparison of otolith biometrics at two locations in the Waters Front of TPI Dudepo Kab. South Bolaang Mongondow and in the Lembeh Strait, Bitung City. Data collection in the field using the roaming survey method. Fishing is done by snorkelling, catching a minimum of 50 individuals representing each size class using Chang net/Sibu fishing gear. The differences in otoliths in male and female sexes have seven different otolith characters in otolith length (OL), otolith width (OW), otolith area (OA) and otolith perimeter (OP), roundness (Rnd), ellipticity (Ell) and aspect ratio (Ar), and three characters that do not show significant differences, namely form factor (Ff), circularity (Cir), rectangularity (Rec). The differences in otoliths based on location in front of the Dudepo TPI and in the Lembeh Strait have seven different otolith characters in otolith length (OL), otolith width (OW), otolith area (OA) and otolith perimeter (OP), roundness (Rnd), ellipticity (Ell). and aspect ratio (Ar), and three characters that do not show significant differences, namely form factor (Ff), circularity (Cir), rectangularity (Rec).
 
 Keywords: Biometrics, cardinal proud fish, comparison
 
 Abstrak
 Penelitian ini bertujuan untuk mengetahui perbandingkan biometrik otolit ikan kardinal banggai berdasarkan jenis kelamin dan mengetahui perbandingan biometrik otolit pada dua lokasi di Perairan Depan TPI Dudepo Kab. Bolaang Mongondow Selatan dan di Selat Lembeh Kota Bitung. Pengambilan data di lapangan menggunakan metode survei jelajah. Penangkapan ikan dilakukan dengan penyelaman snorkling, penangkapan minimal 50 individu yang mewakili setiap kelas ukuran dengan alat tangkap Chang net/Sibu. Perbedaan otolit pada jenis kelamin jantan dan jenis kelamin betina memiliki tujuh perbedaan karakter otolit pada panjang otolit (OL), lebar otolit (OW), area otolit (OA) dan perimeter otolit (OP), roundness (Rnd), ellipticity (Ell) dan aspect ratio (Ar), dan tiga karakter yang tidak menunjukkan perbedaan yang signifikan, yaitu form factor (Ff), circularity (Cir), rectangularity (Rec). Perbedaan otolit berdasarkan lokasi di depan TPI Dudepo dan di Selat Lembeh memiliki tujuh perbedaan karakter otolit pada panjang otolit (OL), lebar otolit (OW), area otolit (OA) dan perimeter otolit (OP), roundness (Rnd), ellipticity (Ell) dan aspect ratio (Ar), dan tiga karakter yang tidak menunjukkan perbedaan yang signifikan, yaitu form factor (Ff), circularity (Cir), rectangularity (Rec).
 
 Kata kunci: Biometrik, ikan banggai kardinal, perbandingan

A recent study was performed to assess the morphometric variation of otoliths of the Decapterus kurroides collected in the northern Sulu and southern Sibuyan seas in delineating fish stocks. Seven of the otolith morphometric descriptors (otolith length, OL; otolith height, OH; otolith weight OW; otolith area, OA; otolith perimeter, OP; ellipticity, EL; and aspect ratio, AR) demonstrated a significant positive correlation with fish length and six (rectangularity, RE; squareness, SQ; roundness, RO; circularity, CI; compactness, CO; and form factor, FF) demonstrated a significant negative correlation. In terms of intraspecific comparisons, almost all the otolith morphometric descriptors were significantly different between the two sites (except RE and OH). Further, principal component (PC) analysis showed that direct otolith morphometric descriptors such as OL, OH, OW, OA, and OP explained 61.71% of the differences (PC1). In contrast, derived otolith descriptors such as EL, AR, and CO explained 22.00% of the variations (PC2) for a total of 83.71% explained variations with the eight otolith morphometric descriptors. Statistics reveal that D. kurroides from the Sibuyan Sea have significantly larger, heavier, and more elliptical otoliths than those from the Sulu Sea. Results suggested that the D. kurroides from Sulu and Sibuyan seas are two different phenotypic stocks. Further studies such as otolith microchemistry, genetics, other life history-based studies, and present exploitation levels should be carried out to supplement the results of this study to fully establish the D. kurroides stock structures within Philippine waters.

In this study, otolith shape indices and relationships between otolith shape and otolith length of the bluefish species Pomatomus saltatrix (Linnaeus, 1766), which was sampled from the Samsun offshore of the Blacksea Region were investigated. A total of 166 samples (94♀ and 72♂) were taken from commercial fishermen in two different fishing seasons. Sagittal otoliths of each sample were removed and six different shape index (form factor, roundness, circularity, rectengularity, ellipticity and aspect ratio) were used in analyses. The linear model is preferred for calculating the relationships between otolith shape properties and otolith length. Paired t-test, wilcoxon test, independent t test were used for statistical analysis. The minimum and maximum total lengths and weights of the captured samples are between 13.5-24.8 cm and 22.01-161.19 g, respectively. There is no difference between female and male in terms of total length and weight. When otolith dimensions of female and male were compared, there were no differences in terms of Otolith Length, Otolith Breadth, Otolith Perimeter and Otolith Area (P˃0.05). However, according to left and right otoliths comparisons, there were differences in terms of OB and OP (P˂0.05). Ellipticity was found to have a much stronger relationship with otolith length than the other five parameters (r2˃0.590).

Otolith shape is species-specific in fish. In this study, 161 samples of Trumpeter sillago Sillago maculata and 164 of Silver sillago Sillago sihama were collected from Beibu Gulf in July 2009. The main objective of this study was to use morphological variables and Fourier harmonics to describe otolith characteristics and use discriminant analyses to separate the two species. Otoliths were measured by traditional one and two dimensional measures (otolith length, width, diameters, area, perimeter, rectangularity and circularity), as well as by Fourier analysis to capture the finer regions of the otolith. Analysis of covariance (ANCOVA) showed that there was significant correlation between morphological variables (diameter, perimeter, otolith length, otolith width, otolith area, density, harmonics 3, harmonics 4, harmonics 5, harmonics 13, harmonics 15, and harmonics 16) and body length. To minimize size effects on the morphological variables between species, only fish with a body length between 90 and 140 mm were included in the data analysis and the variables which had significant relation with body length were transformed using the residual. The result showed that the accuracy of discriminant analysis was 97.8% and 100% for Trumpeter sillago and Silver sillago, respectively. This high accuracy indicated that otolith shape was described accurately by morphological variables and Fourier harmonics, and discriminant analysis was an effective way to identify and separate the two species.

Neritic tunas of the genus Auxis have been commercially exploited as early as the 1970s in the Philippines. However, basic biological information on local stocks remains sparse despite being threatened by population declines. This study investigated the otolith shape morphology and phenotypic stocks of bullet tuna (BT, Auxis rochei) and frigate tuna (FT, A. thazard) within the Celebes Sea (western Celebes Sea, Davao Gulf, Moro Gulf, and Sarangani Bay) using otolith shape analysis. A total of 265 BT and 315 FT otoliths were successfully extracted, and results revealed that as fish grow in body length, otoliths develop more circular and elliptical shapes. All shape morphometric size parameters (otolith length, OL; otolith width, OW; otolith area, OA; and otolith perimeter, OP), alongside shape indices (circularity, CR; ellipticity, EL; rectangularity, RE; form factor, FF; and roundness, RO) showed significant correlations to species’ fork length (FL). Applying the “scaling effect” using linear regression minimizes the allometric growth effects on otoliths. While PERMANOVA successfully delineated phenotypic variations within BT (F = 13.2; p > 0.001) and FT (F = 9.24; p > 0.001) among sites, the Principal Component Analyses also support that all shape indices (CR, EL, FF, RE, RO) of BT and FT contributed most to the stock differentiation. These results imply that both species have distinct spatial populations, which could be attributed to their life-history traits and long-term habitat exposure. This study provides insights into these species’ complex population structures within the Celebes Sea for strengthened site-specific conservation and management measures.

Asteriscus otolith shapes as well as their morphometry and shape contours were investigated in order to identify four allopatric Alburnus species: A. chalcoides (Güldenstädt, 1772) (Ordu), A. escherichii Steindachner, 1897 (Eskişehir), A. mossulensis Heckel, 1843 (Tunceli), and A. tarichi (Güldenstädt, 1814) (Van) in Turkish inland waters. These were compared using the shape indices (form factor, roundness, circularity, ellipticity, rectangularity and aspect ratio), and the morphological characters [otolith weight (OWE), otolith length (OL), otolith width (OW), otolith perimeter (OP), and otolith area (OA)]. The overall canonical discriminant analysis (CDA) classification score was 93.8%, with the lowest score for A. escherichii (82.5%) and the highest for A. chalcoides (100%). The otolith shapes, morphology and shape contours of all sampled fish were a clear species differentiator, thereby demonstrating that the otolith shape is species-specific. The current study presents for the first time comprehensive variation information on interspecific left-right asteriscus otoliths in males and females of each Alburnus species: A. chalcoides from Ordu, A. escherichii from Eskişehir, A. mossulensis from Tunceli and A. tarichi from Van, based on a total of 307 individuals. Scanning electron microscopy (SEM) images, shape contours and other otolith characters vary within the same genus; these differences should be investigated not only in other freshwater fish species or genera but also in the same species living in different habitats. In addition, further investigation is required not only with respect to the morphometry, biometry, shape, geometry, and shape contours of the otoliths, but also regarding the genetic methods for robust identification of various sympatric and allopatric fish populations.

To study the effectiveness of otolith morphological techniques on the population classification of Antarctic ice fish, based on the collected samples of <italic>Champsocephalus gunnari</italic> and <italic>Cryodraco antarcticus</italic>, 5 kinds of basic otolith morphological parameters were measured and converted into 7 morphological indicators, the differences in otolith morphology of two species were compared and analyzed, principal component analysis (PCA), stepwise discriminant analysis (SDA) and random forest (RF) were then used to discriminate the 77 elliptical Fourier harmonics of otolith morphological parameters. The results showed that the otolith parameters and body length of two species were significantly correlated (<italic>P</italic>>0.05). The body length of <italic>C</italic>. <italic>gunnari</italic> showed a power function relationship with some morphological parameters. While otolith length, otolith height, otolith area, otolith perimeter, otolith mass of <italic>C</italic>. <italic>antarcticus</italic> were all in logarithmic relationship with the body length. It could be seen that the otolith ring rate of <italic>C</italic>. <italic>gunnari</italic> was lower than that of <italic>C</italic>. <italic>antarcticus</italic>, compared to the otolith of <italic>C</italic>. <italic>antarcticus</italic>, the otolith of <italic>C</italic>. <italic>gunnari</italic> was rounder and more regular. The conventional morphological analysis could not discriminate the otoliths of two species. The PCA could discriminate the otoliths of two species, and the discriminant success rates of SDA and RF were 98.78% and 91.67%, respectively. SDA has its advantage in small sample size, but RF is suitable for large sample size.

The present study provides the otolith morphology and morphometric relationships with fish size of four deep-sea fishes ( Parascombrops pellucidus Alcock, 1889, Alepocephalus blanfordii Alcock, 1892, Lamprogrammus niger Alcock, 1891, Pterygotrigla hemisticta (Temminck &amp; Schlegel, 1843)) collected from western Bay of Bengal during March 2020. Among these, the equations were derived for the first time for three species ( P. pellucidus , A. blanfordii , L. niger ). Sampling was done as a part of deep-sea exploratory survey of FORV Sagar Sampada along the deeper shelf regions of the Bay of Bengal at a depth range of 200 – 1000 m using high-speed demersal trawl (HSDT-CV). The numerical relations established using regression between fish size (TL) and various otolith morphometric measurements (otolith length (OL), otolith height (OH), otolith weight (OWe), otolith area (OA) and otolith perimeter (OP)) can be used to predict the prey size in food and feeding studies for studying the food web dynamics of less-studied deep-sea fishes. LWR of the otolith of selected species showed a negative allometric growth ( t- test, p &lt; 0.5). The higher r 2 value ( &gt; 0.70) obtained for the relationship between fish size (TL) and various otolith morphometric measurements indicates the robustness of the model. The representative images of otoliths of these fishes will be helpful to the taxonomists for the species confirmation and reconstruction of past species assemblages in the palaeontological studies.

Using a radiomics framework to quantitatively analyze tumor shape and texture features in three dimensions, we tested its ability to objectively and robustly distinguish between benign and malignant renal masses. We assessed the relative contributions of shape and texture metrics separately and together in the prediction model. Computed tomography (CT) images of 735 patients with 539 malignant and 196 benign masses were segmented in this retrospective study. Thirty-three shape and 760 texture metrics were calculated per tumor. Tumor classification models using shape, texture, and both metrics were built using random forest and AdaBoost with tenfold cross-validation. Sensitivity analyses on five sub-cohorts with respect to the acquisition phase were conducted. Additional sensitivity analyses after multiple imputation were also conducted. Model performance was assessed using AUC. Random forest classifier showed shape metrics featuring within the top 10% performing metrics regardless of phase, attaining the highest variable importance in the corticomedullary phase. Convex hull perimeter ratio is a consistently high-performing shape feature. Shape metrics alone achieved an AUC ranging 0.64-0.68 across multiple classifiers, compared with 0.67-0.75 and 0.68-0.75 achieved by texture-only and combined models, respectively. Shape metrics alone attain high prediction performance and high variable importance in the combined model, while being independent of the acquisition phase (unlike texture). Shape analysis therefore should not be overlooked in its potential to distinguish benign from malignant tumors, and future radiomics platforms powered by machine learning should harness both shape and texture metrics. • Current radiomics research is heavily weighted towards texture analysis, but quantitative shape metrics should not be ignored in their potential to distinguish benign from malignant renal tumors. • Shape metrics alone can attain high prediction performance and demonstrate high variable importance in the combined shape and texture radiomics model. • Any future radiomics platform powered by machine learning should harness both shape and texture metrics, especially since tumor shape (unlike texture) is independent of the acquisition phase and more robust from the imaging variations.

Fish otoliths are conservative structures that are widely used on fishery science for multiple purposes. Despite its relevance in the research field, little is known about the ontogeny and inter-population effects on the otolith of freshwater fish. In this study we used otoliths from 1800 European perch (Perca fluviatilis) individuals from 9 different populations to analyze the ontogenetic and inter-population differences on the otolith shape using six morphometric indices. The relationship between fish and otolith length was fitted using three different regression models (linear, power and logistic) to identify the best allometric relationship. Our results show that there are strong ontogenetic and interpopulation differences in P. fluviatilis otolith shapes. We also show that the relationship between the fish and otolith length follows a logistic curve. The ontogenetic differences on otolith shape might be related to extrinsic factors (diet shift and intra and interspecific competition) in each ecosystem, given that the reservoirs are different and no clear pattern on the otolith shapes can be distinguished among populations. Our results imply that the available back-calculation models may not always provide accurate estimates of P. fluviatilis length and that a model that takes into account the real allometric relationship for the species can improve the fish length estimations of back-calculated fish lengths for the European perch.

In this study, relationships between fish length and otolith length, width and weight of Sardinella sindensis from Bandar Lengeh and Qeshm Island, Persian Gulf were analyzed. In total, 128 and 120 fishes collected from Commercial catches during March 2011-February 2012 in the Bandar Lengeh and Qeshm Island repectively. There were no significant differences between left and right otolith (t-test, P>0.05) and between males and females otolith (ANCOVA, P>0.05). For these reasons, only right otoliths were used for next analysis and data of each two sex were pooled. Relationships between length and otolith length, width and weight described by linear regression models and high correlation showed for all relationships. The highest correlation was between fish length and otolith length (Bandar Lengeh, R2=0.8722; Qeshm Island, R2=0.8661) and relashionship between fish length and otolith width was less correlation than other relathionships (Bandar Lengeh, R2=0.7355; Qeshm Island, R2=0.7275). These results showed that fish length and otolith growth have a positive relashionship, so can be useful tools for determining fish species, size and age.

Sardines are the most important commercial fishes for Iran. However, information about biology and ecology of sind sardine in Persian Gulf and Oman Sea is scarce. In this study, relationship between fish length and otolith length, -width and -weight of Sardinella sindensis from Bandar Lengeh and Qeshm Island, Persian Gulf were analyzed. In total, 128 and 120 fishes collected from Commercial catches during March 2011- February 2012 in the Bandar Lengeh and Qeshm Island respectively. There were no significant differences between left and right otolith (t-test, P > 0.05) or between males and females otoliths (ANCOVA, P > 0.05). For these reasons, only right otoliths were used for next analysis and data of both sexes were pooled. Relationships between fish length and otolith length, width and weight were described by linear regression models and high correlation was shown for all relationships. The highest correlation was between fish length and otolith length (Bandar Lengeh, R 2 = 0.8722; Qeshm Island, R 2 = 0.8661) and relationship between fish length and otolith width showed less correlation than other relationships (Bandar Lengeh, R 2 = 0.7355; Qeshm Island, R 2 = 0.7275). The result from this study shows that fish length and otolith growth have a positive relationship, so can be a useful tool to evaluate the growth of fish and fish stocks and finally fisheries management.

The present study was conducted to investigate the relationships between fish length, otolith size and otolith weight. A total of 180 Sperata aor and 171 Labeo bata specimens were collected monthly during January 2016 to March 2017 from the River Ganga at the Narora site. Student’s t-test showed no significant differences in the size of right and left otoliths in the selected fish species, therefore, a single linear regression based on the left otolith was carried out. Fish length was plotted against otolith weight, otolith length and otolith height. Moreover, otolith length was plotted against otolith weight and otolith height. The linear regression model was found to fit the data well for fish length to otolith size in both selected fish species. A strong relationship between otolith length and weight was found in both species (r2 &gt; 0.8). Furthermore, the study revealed that fish length is potentially related to otolith size and a strong relationship between fish length and otolith size exists in S. aor (mean r2 = 0.89) as compared to L. bata (mean r2 = 0.70). Such relationships should be determined for the fish species that are ecologically and socio-economically important in order to understand their trophic relationships, population dynamics and yield estimates.

Relationships between fish length and otolith length and width were examined in the blueline snapper Lutjanus coeruleolineatus (Lutjanidae) collected from the coasts of Oman, Arabian Sea. The values of exponent b from the relationships between fish total length and otolith length and total length and otolith width were estimated representing the close fitness of otolith size with fish size. Both relationships were statistically significant, which means both otolith length and width can be used to retrieve the fish original size. The analysis of covariance (ANCOVA) was used to test the effect of the categorical factor of species in the fish length and otolith length relationship. This study represents the first reference available on the relationship of fish size and otolith size for L. coeruleolineatus in the Arabian Sea area. Results from this study will offer original data on quantitative biometric relationships between body and otolith measurements of fish species in Arabian Sea region. By obtaining the mathematical model showing the relationship of the otolith size and fish length will enable fisheries biologists to know the size of the fish that has been eaten an information which important for fish biologists.

Fifty-four specimens of the kingsoldier bream, Argyrops spinifer (Forsskal, 1775) from coastal waters of the Persian Gulf, in Bushehr Province, were aged by otolith readings to assess the suitability of using otolith length/weight for age estimation in this fish. Otolith length and weight increased along with an increase in length and age of the fish. The relationship between fish length (TL) and otolith weight (OW) was as OW=0.007TL –0.79 (r 2 =0.90), between fish age (t) and otolith length (OL) as OL=5.576 (t) 0.346 (r 2 =0.88), between fish age (t) and otolith weight (OW) as OW=0.03 (t) 0.837 (r 2 =0.91) and between fish length (TL) and otolith length (OL) as OL=1.125 (TL) 0.644 (r 2 =0.92). Thus, these parameters could be readily used in age determination of A. spinifer. The proportion of average otolith weight to fish total length which partly shows the growth rate, was assessed as 0.005.