Response to Packard: make sure we do not throw out the biological baby with the statistical bath water when performing allometric analyses

A Revision of the Traditional Analysis Method of Allometry to Allow Extension of the Normality-Borne Complexity of Error Structure: Examining the Adequacy of a Normal-Mixture Distribution-Driven Error Term

Huxley’s model of simple allometry provides a parsimonious scheme for examining scaling relationships in scientific research, resource management, and species conservation endeavors. Factors including biological error, analysis method, sample size, and overall data quality can undermine the reliability of a fit of Huxley’s model. Customary amendments enhance the complexity of the power function-conveyed systematic term while keeping the usual normality-borne error structure. The resulting protocols bear multiple-parameter complex allometry forms that could pose interpretative shortcomings and parameter estimation difficulties, and even being empirically pertinent, they could potentially bear overfitting. A subsequent heavy-tailed Q-Q normal spread often remains undetected since the adequacy of a normally distributed error term remains unexplored. Previously, we promoted the advantages of keeping Huxley’s model-driven systematic part while switching to a logistically distributed error term to improve fit quality. Here, we analyzed eelgrass leaf biomass and area data exhibiting a marked size-related heterogeneity, perhaps explaining a lack of systematization at data gathering. Overdispersion precluded adequacy of the logistically adapted protocol, thereby suggesting processing data through a median absolute deviation scheme aimed to remove unduly replicates. Nevertheless, achieving regularity to Huxley’s power function-like trend required the removal of many replicates, thereby questioning the integrity of a data cleaning approach. But, we managed to adapt the complexity of the error term to reliably identify Huxley’s model-like systematic part masked by variability in data. Achieving this relied on an error term conforming to a normal mixture distribution which successfully managed overdispersion in data. Compared to normal-complex allometry and data cleaning composites present arrangement delivered a coherent Q-Q normal mixture spread and a remarkable reproducibility strength of derived proxies. By keeping the analysis within Huxley’s original theory, the present approach enables substantiating nondestructive allometric proxies aimed at eelgrass conservation. The viewpoint endorsed here could also make data cleaning unnecessary.

Biological scaling analyses are more than statistical line fitting

The magnitude of many biological traits relates strongly and regularly to body size. Consequently, a major goal of comparative biology is to understand and apply these 'size-scaling' relationships, traditionally quantified by using linear regression analyses based on log-transformed data. However, recently some investigators have questioned this traditional method, arguing that linear or non-linear regression based on untransformed arithmetic data may provide better statistical fits than log-linear analyses. Furthermore, they advocate the replacement of the traditional method by alternative specific methods on a case-by-case basis, based simply on best-fit criteria. Here, I argue that the use of logarithms in scaling analyses presents multiple valuable advantages, both statistical and conceptual. Most importantly, log-transformation allows biologically meaningful, properly scaled (scale-independent) comparisons of organisms of different size, whereas non-scaled (scale-dependent) analyses based on untransformed arithmetic data do not. Additionally, log-based analyses can readily reveal biologically and theoretically relevant discontinuities in scale invariance during developmental or evolutionary increases in body size that are not shown by linear or non-linear arithmetic analyses. In this way, log-transformation advances our understanding of biological scaling conceptually, not just statistically. I hope that my Commentary helps students, non-specialists and other interested readers to understand the general benefits of using log-transformed data in size-scaling analyses, and stimulates advocates of arithmetic analyses to show how they may improve our understanding of scaling conceptually, not just statistically.

Back to the basics: allometric growth by the horns of bovid mammals

I used the equivalent of nonlinear analysis of covariance (ANCOVA) to re-examine relative growth by the horns on males and females of alpine ibex (Capra ibex) and mouflon sheep (Ovis gmelini). A prior study of allometric growth by the horns on these animals described a pattern of biphasic allometry for both sexes, with two different mathematical equations being required to capture the pattern of variation over the full range in body size. However, the investigation in question used conventional analytical methods based on logarithmic transformations, which alter bivariate distributions and commonly introduce problems with analysis and interpretation. My new analyses of data for both species revealed that untransformed observations for both males and females are monophasic and that they are described quite well by three-parameter power equations with negative intercepts. Equations for males follow a steep upward trajectory whereas those for females follow much shallower paths. The negative intercepts indicate that males and females of both species must attain a minimum body size before horns begin to develop. Conclusions from the earlier investigation were based on inaccurate perceptions of pattern in the data. Future studies should be based on graphical and analytical analysis of observations expressed on the original arithmetic scale.

Measuring Complex Morphological Traits with 3D Photogrammetry: A Case Study with Deer Antlers

The increasing availability of 3D-imaging technology provides new opportunities for measuring morphology. Photogrammetry enables easy 3D-data acquisition compared to conventional methods and here we assess its accuracy for measuring the size of deer antlers, a complex morphological structure. Using a proprietary photogrammetry software, we generated 3D images of antlers for 92 individuals from 29 species of cervids that vary widely in antler size and shape and used these to measure antler volume. By repeating the process, we found that the relative error averaged 8.5% of object size. Errors in converting arbitrary voxel units into real volumetric units accounted for 70% of the measurement variance and can therefore be reduced by replicating the conversion. We applied the method to clay models of known volume and found no indication of bias. The estimation was robust against variation in imaging device, distance and operator, but approximately 40 images per specimen were necessary to achieve good precision. We used the method to show that conventional measures of main-beam length are relatively poor estimators of antler volume. Using loose antlers of known weight, we also showed that the volume may be a relatively poor predictor of antler weight due to variation in bone density across species. We conclude that photogrammetry can be an efficient and accurate tool for measuring antlers, and likely many other complex morphological traits.

Assessment of a Takagi-Sugeno-Kang fuzzy model assembly for examination of polyphasic loglinear allometry

Background

The traditional allometric analysis relies on log- transformation to contemplate linear regression in geometrical space then retransforming to get Huxley’s model of simple allometry. Views assert this induces bias endorsing multi-parameter complex allometry forms and nonlinear regression in arithmetical scales. Defenders of traditional approach deem it necessary since generally organismal growth is essentially multiplicative. Then keeping allometry as originally envisioned by Huxley requires a paradigm of polyphasic loglinear allometry. A Takagi-Sugeno-Kang fuzzy model assembles a mixture of weighted sub models. This allows direct identification of break points for transition between phases. Then, this paradigm is seamlessly appropriate for efficient allometric examination of polyphasic loglinear allometry patterns. Here, we explore its suitability.

Methods

Present fuzzy model embraces firing strength weights from Gaussian membership functions and linear consequents. Weights are identified by subtractive clustering and consequents through recursive least squares or maximum likelihood. Intersection of firing strength factors set criterion to estimate breakpoints. A multi-parameter complex allometry model follows by adapting firing strengths by composite membership functions and linear consequents in arithmetical space.

Results

Takagi-Sugeno-Kang surrogates adapted complexity depending on analyzed data set. Retransformation results conveyed reproducibility strength of similar proxies identified in arithmetical space. Breakpoints were straightforwardly identified. Retransformed form implies complex allometry as a generalization of Huxley’s power model involving covariate depending parameters. Huxley reported a breakpoint in the log–log plot of chela mass vs. body mass of fiddler crabs (Uca pugnax), attributed to a sudden change in relative growth of the chela approximately when crabs reach sexual maturity. G.C. Packard implied this breakpoint as putative. However, according to present fuzzy methods existence of a break point in Huxley’s data could be validated.

Conclusions

Offered scheme bears reliable analysis of zero intercept allometries based on geometrical space protocols. Endorsed affine structure accommodates either polyphasic or simple allometry if whatever turns required. Interpretation of break points characterizing heterogeneity is intuitive. Analysis can be achieved in an interactive way. This could not have been obtained by relying on customary approaches. Besides, identification of break points in arithmetical scale is straightforward. Present Takagi-Sugeno-Kang arrangement offers a way to overcome the controversy between a school considering a log-transformation necessary and their critics claiming that consistent results can be only obtained through complex allometry models fitted by direct nonlinear regression in the original scales.

Biological interpretations of the biphasic model of ontogenetic brain–body allometry: a reply to Packard

Allometry is a description of organismal growth. Historically, a simple power law has been used most widely to describe the rate of growth in phenotypic traits relative to the rate of growth in overall size. However, the validity of this standard practice has repeatedly been criticized. In an accompanying opinion piece, Packard reanalysed data from a recent study on brain–body ontogenetic allometry and claimed that the biphasic growth model suggested in that study was an artefact of logarithmic transformation. Based on the model selection, Packard proposed alternative hypotheses for brain–body ontogenetic allometry. Here, I examine the validity of these models by comparing empirical data on body sizes at two critical neurodevelopmental events in mammals, i.e. at birth and at the time of the peak rate of brain growth, with statistically inferred body sizes that are supposed to characterize neurodevelopmental processes. These analyses support the existence of two distinct phases of brain growth and provide weak support for Packard's uniphasic model of brain growth. This study demonstrates the importance of considering alternative models in studies of allometry, but also highlights that such models need to respect the biological theoretical context of allometry.

Examination of the Effects of Curvature in Geometrical Space on Accuracy of Scaling Derived Projections of Plant Biomass Units: Applications to the Assessment of Average Leaf Biomass in Eelgrass Shoots

Conservation of eelgrass relies on transplants and evaluation of success depends on nondestructive measurements of average leaf biomass in shoots among other variables. Allometric proxies offer a convenient way to assessments. Identifying surrogates via log transformation and linear regression can set biased results. Views conceive this approach to be meaningful, asserting that curvature in geometrical space explains bias. Inappropriateness of correction factor of retransformation bias could also explain inconsistencies. Accounting for nonlinearity of the log transformed response relied on a generalized allometric model. Scaling parameters depend continuously on the descriptor. Joining correction factor is conceived as the partial sum of series expansion of mean retransformed residuals leading to highest reproducibility strength. Fits of particular characterizations of the generalized curvature model conveyed outstanding reproducibility of average eelgrass leaf biomass in shoots. Although nonlinear heteroscedastic regression resulted also to be suitable, only log transformation approaches can unmask a size related differentiation in growth form of the leaf. Generally, whenever structure of regression error is undetermined, choosing a suitable form of retransformation correction factor becomes elusive. Compared to customary nonparametric characterizations of this correction factor, present form proved more efficient. We expect that offered generalized allometric model along with proposed correction factor form provides a suitable analytical arrangement for the general settings of allometric examination.

Scaling of work and power in a locomotor muscle of a frog

Muscle work and power are important determinants of movement performance in animals. How these muscle properties scale determines, in part, the scaling of performance during movements, such as jump height or distance. Muscle-mass-specific work is predicted to remain constant across a range of scales, assuming geometric similarity, while muscle-mass-specific power is expected to decrease with increasing scale. We tested these predictions by examining muscle morphology and contractile properties of plantaris muscles from frogs ranging in mass from 1.28 to 20.60 g. Scaling of muscle work and power was examined using both linear regression on log₁₀-transformed data (LR) and non-linear regressions on untransformed data (NLR). Results depended on the method of regression not because of large changes in scaling slopes, but because of changing levels of statistical significance using corrections for multiple tests, demonstrating the importance of careful consideration of statistical methods when analyzing patterns of scaling. In LR, muscle-mass-specific work decreased with increasing scale, but an accompanying positive allometry of muscle mass predicts constant movement performance at all scales. These relationships were non-significant in NLR, though scaling with geometric similarity also predicts constant jump performance across scales, because of proportional increases in available muscle energy and body mass. Both intrinsic shortening velocity and muscle-mass-specific power were positively allometric in both types of analysis. Nonetheless, scale accounts for little variation in contractile properties overall over the range examined, indicating that other sources of intraspecific variation may be more important in determining muscle performance and its effects on movement.