Effects of stomach content and gonad mass on length–weight relationships and condition indices in fish species with different body shapes

This study was conducted in October 2020 in Güllük Bay, located in the southern Aegean Sea, Türkiye, at depths ranging from 40 m to 80 m (Figure 1). Fish samples were collected during a single survey using a traditional commercial bottom trawl net with a stretched mesh size of 44 mm.

Figure 1

An overview of the study area.

To examine whether the influence of internal organ mass differs among body forms, three species representing contrasting body shapes were selected: B. boops (fusiform), C. linguatula (flat), and D. annularis (laterally compressed). Individuals were randomly subsampled from the catch and transported on ice to the laboratory for morphometric and gravimetric analyses. Sample sizes reflected the number of suitable individuals available from the standardized survey catch and were considered sufficient for comparative exploratory analyses across weight scenarios.

Because sampling was conducted outside the main reproductive season of the studied species, the collected specimens did not represent peak gonadal development. Therefore, the empirical dataset did not include naturally occurring maximum gonad masses. To evaluate the potential influence of reproductive investment on LWRs and Fulton’s condition factor, a simulated gonad-weight scenario was constructed using peak GSI values reported in the literature for each species.

In the laboratory, the total length (TL) of each fish was measured to the nearest 0.1 cm, and body weight was recorded to the nearest 0.01 g using a precision balance.

To evaluate the effects of internal organ mass and reproductive investment on weight-based metrics, three weight categories were considered for each individual:

Total body weight (TW): measured whole-body weight, including internal organs

Gutted weight (F): body weight after removal of internal organs

Gonad-adjusted weight (TWG): estimated weight obtained by adding a simulated gonad mass to the measured total body weight (TW).

Because the sampled individuals were collected outside the peak reproductive period, the observed gonad masses were not considered representative of maximum reproductive investment. In addition, the sex of individuals was not determined, and analyses were conducted on mixed-sex samples. Because the study focused on generalized weight-scenario comparisons rather than sex-specific reproductive biology, analyses were conducted at the population level. Therefore, gonad effects were evaluated using a simulated approach. Specifically, TWG values were generated by incorporating species-specific peak GSI values obtained from the literature, representing the upper bound of reproductive investment at the population level rather than individual-level variation.

Simulated gonad mass was added to measured total body weight according to the following equation: TWG = TW + (TW×GSI/100) {\rm{TWG = TW + (TW}} \times {\rm{GSI/100)}}

Where GSI represents the species-specific peak reproductive value derived from the literature.

Accordingly, the following peak GSI values were used:

B. boops: 5.0% (Soykan et al., 2015)

C. linguatula: 3.2% (Ulutürk et al., 2016)

D. annularis: 15.4% (Mouine et al., 2012)

These gonad-adjusted weights were not directly measured in the sampled fish and should therefore be interpreted as a simulated reproductive scenario, rather than an observed field condition. Because maturity status was not determined, the simulated GSI values should not be interpreted as biologically attainable values for every individual, particularly for immature specimens.

Stomach fullness was visually assessed for each specimen using six semi-quantitative categories based on the proportion of stomach volume occupied by food contents: empty (0%), trace (12.5%), quarter full (25%), half full (50%), three-quarters full (75%), and full (100%), following El-Maremie and El-Mor (2015). Stomach contents were not weighed directly; therefore, feeding-related mass variation was evaluated only semi-quantitatively.

LWRs were estimated separately for each species and each weight type (F, TW, and TWG) using the traditional power function: W=a× L b W = a \times {L^b}

Where W is body weight (g), L is total length (cm), a is the intercept, and b is the allometric coefficient.

To linearize the relationship, both variables were log₁₀-transformed: log 10 W= log 10 a+b log 10 L {\log _{10}}W = {\log _{10}}a + {\rm{b}}{\log _{10}}L

Linear regression models were fitted for each species and weight category. For each model, the parameters a, b, the standard error of b, the coefficient of determination (R²), and 95% confidence intervals were calculated.

To determine whether the allometric coefficient (b) differed among weight types within each species, an analysis of covariance (ANCOVA) framework was applied using log-transformed data. For each species, the following model was fitted: log 10 W∼ log 10 L×WeightType {\log _{10}}W \sim {\log _{10}}L \times {\rm{WeightType}}

Where WeightType included F, TW, and TWG. The interaction term (log₁₀L × WeightType) was used to test whether slopes differed significantly among weight types. This interaction term was also used to evaluate the homogeneity of regression slopes assumption required for ANCOVA. A non-significant interaction indicated that the estimated b values did not differ among weight categories.

In addition, departure from isometric growth (b = 3) was evaluated for each regression using the t-test described by Morey et al. (2003): T=( b−3 )/Sb T = \left( {b - 3} \right)/{\rm{S}}b

Where Sb is the standard error of the slope.

Because F, TW, and TWG values were derived from the same individuals, all comparisons among weight types were treated as paired measurements. Accordingly, repeated-measures designs were used to account for within-individual dependence among weight scenarios. Pairwise comparisons among weight definitions were conducted using paired tests on log-transformed values because all weight scenarios originated from the same individuals. Normality of paired differences was evaluated visually using diagnostic plots. Since only three repeated weight scenarios were compared per individual, no additional sphericity correction was applied. The homogeneity of regression slopes assumption for ANCOVA was evaluated using the interaction term between log10L and WeightType.

Differences among mean body weights (F, TW, and TWG) within each species were evaluated using repeated-measures analysis. When assumptions for parametric repeated-measures analysis were not satisfied, pairwise paired tests on log-transformed values were preferred.

To further explore the influence of reproductive investment in D. annularis, a sensitivity analysis was conducted by recalculating TWG under incrementally increasing hypothetical GSI values. This analysis was used to identify the approximate GSI range at which differences between measured and gonad-adjusted weights became statistically detectable.

Fulton’s condition factor (K) was calculated for each individual and each weight type using: K=100×( W/ L 3 ) K = 100 \times \left( {W/{L^3}} \right)

Where W is the corresponding weight type (F, TW, or TWG) and L is total length (cm).

Because condition factor values were calculated repeatedly for the same individuals under different weight scenarios, differences among mean K values were analyzed using repeated-measures comparisons within each species.

Data organization was performed in Microsoft Excel, whereas all statistical analyses were conducted in RStudio (Posit team, 2025). Prior to modeling, data were visually inspected and log10-transformed where appropriate to improve linearity and homoscedasticity. Statistical significance was accepted at α = 0.05. Model assumptions, including residual normality, homogeneity of variance, and linearity, were evaluated visually using residual diagnostic plots. Residual diagnostic plots indicated no major deviations from normality, homoscedasticity, or linearity assumptions for the fitted models. Visual inspection of residual-versus-fitted and Q–Q plots did not reveal systematic patterns that would compromise model interpretation. Therefore, no additional model corrections were considered necessary.

The high proportion of empty stomachs observed across all species suggests that short-term feeding contributed minimally to body mass during sampling. This is consistent with previous studies reporting high frequencies of empty stomachs in natural fish populations (Baker et al., 2014; Hyslop, 1980), particularly under conditions of variable prey availability and foraging success (Gerking, 1994). Under such circumstances, stomach content is unlikely to introduce substantial variability in total body weight. Therefore, the low stomach fullness observed in this study suggests that stomach-content mass likely contributed only limited variability to body weight under the present sampling conditions. However, because stomach contents were not weighed directly, the present findings should be interpreted cautiously.

An important outcome of this study is the empirical confirmation that neither internal organ removal (F) nor simulated gonad addition (TWG) substantially altered the allometric coefficient (b) of the LWR across the studied species. Despite clear differences in absolute body weight among F, TW, and TWG, the slopes of the relationships remained statistically indistinguishable.

This result indicates that internal organ mass did not affect the allometric growth pattern of fish in the present study, which is primarily determined by morphological and structural characteristics (Kulbicki et al., 2005). Instead, internal components introduce vertical shifts in the LWR by affecting the intercept (a) while leaving the scaling exponent (b) unchanged (Le Cren, 1951). This suggests that LWR-based growth interpretations are robust to variations in internal organ mass, even under scenarios of increased reproductive investment. Accordingly, the novelty of the present study lies less in demonstrating a previously unknown mathematical property of LWRs than in evaluating the robustness of these relationships under biologically modified weight scenarios across species with contrasting body forms and reproductive investment.

Although LWR slopes remained unchanged, pairwise comparisons of body weights revealed species-specific responses to simulated gonad mass. No significant differences were detected among weight definitions in B. boops and C. linguatula, whereas D. annularis exhibited significant weight increases when gonad mass was incorporated.

This divergence can be explained by differences in reproductive investment among species (Murua & Saborido-Rey, 2003; Wootton, 1998). D. annularis, characterized by a relatively high GSI, appears more sensitive to the inclusion of gonad mass, resulting in measurable increases in total body weight. In contrast, the lower GSI values reported for B. boops and C. linguatula likely remain below the threshold required to produce statistically detectable changes.

Previous studies have demonstrated that gonad weight varies seasonally and can substantially contribute to total body mass (Le Cren, 1951), while the magnitude of this contribution differs widely among species, with GSI values ranging from low levels (~5%) to over 35% (Rijnsdorp et al., 2015). This broad variability supports the interpretation that in species with relatively low GSI, gonad mass may not generate detectable differences in total body weight, whereas in species with higher reproductive investment, its effect becomes more pronounced.

These findings highlight that the impact of gonad mass on weight-based metrics is not universal but depends on species-specific reproductive strategies.

The sensitivity analysis performed for D. annularis further clarifies this pattern by identifying threshold GSI levels at which gonad mass begins to influence statistical outcomes. Within the present simulation framework, differences between F and TWG became statistically significant at approximately 5% GSI, whereas differences between TW and TWG emerged around 13% GSI.

This threshold-based response suggests that reproductive investment must reach a certain magnitude before it can meaningfully affect weight-based comparisons. Similar patterns are consistent with general life-history theory, where energy allocation to reproduction increases at the expense of somatic growth (Stearns, 1992, 2000). Given the wide variability in GSI among fish species (Rijnsdorp et al., 2015), the effect of gonad mass on total body weight is expected to remain negligible at low GSI levels but become increasingly important as reproductive investment increases (Wootton, 1998). In some species, enlarged gonads may also reduce abdominal space available for feeding, potentially limiting stomach fullness and food intake during peak reproductive periods (Weeks, 1996). However, the present simulation assumes that gonad mass increases independently of other body components. In natural reproductive cycles, increasing gonad investment may coincide with reductions in somatic tissues or energy reserves, including liver and muscle mass, potentially moderating the net effect on total body weight. Therefore, the simulated TWG values may represent an upper-bound estimate of gonad-related weight effects. Actual threshold responses are likely to vary among populations, sexes, maturity stages, and environmental conditions.

Importantly, these thresholds should not be interpreted as universal biological reference values, but rather as simulation-based approximations illustrating the potential sensitivity of weight-based metrics to increasing reproductive investment. Species with low GSI values may be analyzed without adjusting for gonad weight, whereas species with high reproductive investment may require more careful standardization of weight measurements (Froese, 2006).

In contrast to LWR results, Fulton’s condition factor (K) was significantly affected by weight definition across all species. Values derived from TWG were consistently higher, while those based on F were lowest. This consistent pattern reflects the direct dependence of K on body weight, making it inherently sensitive to any variation in mass, regardless of its biological origin (Froese, 2006; Le Cren, 1951).

This finding highlights an important limitation of Fulton’s condition factor. Previous studies have shown that condition indices can be influenced by internal biological factors and may introduce bias if not carefully interpreted (Bolger & Connolly, 1989). Therefore, comparisons of condition factor across studies or sampling periods may be misleading if weight definitions are not standardized (Blackwell et al., 2000; Jakob et al., 1996). The discrepancy between the stability of LWR parameters and the sensitivity of K underscores the need for caution when interpreting condition indices. This contrast also highlights the value of evaluating LWR robustness together with condition-based metrics when assessing biologically driven variation in fish body weight.

In addition, Fulton’s condition factor assumes isometric growth (b = 3), and its interpretation may become problematic when the observed LWR deviates substantially from this assumption. Under positive or negative allometric growth, K may vary systematically with body size, meaning that comparisons among individuals of markedly different lengths can be influenced by size-related scaling effects rather than differences in physiological condition alone. In the present study, this limitation is unlikely to affect the main conclusions because comparisons among weight definitions were performed using repeated measurements on the same individuals, thereby controlling for body size. Nevertheless, the observed differences in b values among species may partly reflect differences in body shape and growth form, highlighting the importance of considering species-specific length–weight scaling when interpreting condition indices based on body weight.

The contrasting responses observed among species may reflect differences in reproductive investment and body morphology; however, the relative contribution of body shape alone remains uncertain. Although D. annularis exhibited a stronger response to simulated gonad mass, this pattern cannot be attributed solely to body shape because reproductive investment and condition dynamics vary substantially among species independent of morphology (Wuenschel et al., 2019). Previous studies have demonstrated that reproductive state, maturity, and energy-allocation strategy may strongly influence condition-related metrics across species with similar body forms. Therefore, the present findings should be interpreted with caution, and additional comparative studies including species with contrasting GSI values and body forms would be necessary to isolate the specific role of morphology.

These observations are consistent with previous studies highlighting the role of body shape in determining weight–length relationships and biomass distribution patterns (Kulbicki et al., 2005). Moreover, morphological differences among species are known to influence energy allocation and internal tissue organization (McBride et al., 2015; Webb, 1984). Although this hypothesis requires further investigation, it offers a plausible biological explanation that links morphology, reproductive investment, and weight-based metrics.

This study used a simulation-based approach to represent peak reproductive investment, as sampling was conducted outside the main spawning period. While this approach enables controlled evaluation of gonad effects, it represents a theoretical maximum rather than observed field conditions.

The simulated gonad-adjusted weights used in this study may slightly overestimate the effect of reproductive investment because the measured total body weight (TW) already included the gonad mass present at the time of sampling. Since gonads were not weighed separately, the existing gonadal contribution could not be removed prior to simulation. Consequently, the simulated procedure effectively added peak GSI-derived gonad mass on top of an unknown baseline gonad mass, potentially inflating the estimated weight increase. However, sampling was conducted outside the main reproductive season, when gonad development is expected to be relatively limited, suggesting that this bias is likely small. Future studies incorporating direct measurements of gonad weight would allow replacement of observed gonadal mass with simulated reproductive scenarios and provide more accurate estimates of gonad-related effects on weight-based metrics.

In addition, analyses were conducted on mixed-sex samples without distinguishing between males and females. Given that gonad development can differ substantially between sexes, the use of species-level GSI values may mask individual variability. However, the approach remains valid for assessing the upper-bound influence of reproductive investment on weight-based metrics. Accordingly, the simulated gonad-adjusted weights should be interpreted as a theoretical sensitivity scenario representing population-level reproductive potential rather than actual gonad masses of the sampled individuals.