REVIEW 4 major objections 5 minor 15 references
Quantifying movement: Expanding the Ichnologist toolkit
T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash
Pith's one-line read The paper claims that fossil movement paths can be discretized into width-normalized steps and turning angles, and that this quantitative method reveals three recurring behavioural morphotypes in Cruziana semiplicata that persist across loc
desk verdict The discretization toolkit is a real contribution; the three-morphotype persistence claim is built on circular t-test logic and needs rework before it can be believed. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is the turning-angle distribution of a discretized path. A trail is traced from photographs into coordinates, resampled at intervals equal to a chosen multiple of the trail width (here 0.5 times width), and each triple of consecutive points is converted into one turning angle. Each specimen, subgroup, or locality is then represented by the distribution of these angles, and pairs of distributions are compared with Welch's two-sample t-tests; the resulting p-value matrices are clustered by eye into morphotypes. The turning-angle distribution carries the argument because it summarizes path morphology in a form that can be tested statistically, while the width-relative segment
What would settle it
Re-analyse the p-value matrix with a multiple-comparison correction (e.g., Benjamini–Hochberg) and compute the statistical power of the pairwise t-tests under the observed variances (Russia, for example, has variance 15.20 with 12 paths). If the checkerboard clustering into morphotypes 2 and 3 does not survive correction, or if power is below conventional levels for the group comparisons, the three-morphotype persistence claim is not established.
Extended reading notes
Core claim
The central claim is that fossil movement paths carry quantifiable behavioural signal once they are discretized into width-relative steps and summarized as turning-angle distributions. On Cruziana semiplicata, this signal takes the form of three morphotypes: one dominant stereotyped movement pattern and two subordinate behavioural variants. The dominant pattern is statistically similar before and after resting traces, and it persists across five geographically separate localities once the subordinate specimens are set aside. The authors therefore conclude that locality-level behavioural differences are a sampling artefact of rare morphotypes, and that the proposed two-ichnosubspecies split i
Load-bearing premise
The argument relies on treating p > 0.1 from small-sample t-tests as evidence that two movement patterns are the same or that behaviour persists; with as few as three to five paths in some localities and no power analysis, a non-significant p-value is weak support for that conclusion.
Editorial extensions
If this is right
- Ichnologists can treat trail and trackway specimens as movement datasets and apply the statistical toolkit of movement ecology to fossil behaviour.
- For C. semiplicata, the path after a resting trace is statistically indistinguishable from the path before it, supporting the view of a fixed behavioural programme.
- The dominant morphotype is shared across Spain, Oman, Poland, Russia, and Wales once rare subordinate specimens are set aside, so locality-level differences are a sampling effect.
- The proposed two-ichnosubspecies split of C. semiplicata is replaced by three morphotypes—one common, two subordinate—that occur in varying proportions.
- Because the output measures are unitless and width-relative, the same workflow can be applied across different ichnospecies, environments, and geological time intervals.
Reading between the lines
- Because segment length is tied to trail width, the method implicitly assumes movement speed scales with body size; if that scaling is wrong, the inferred 'time' axis and hence the turning-angle comparison would be biased. Testing this on modern tracemakers of different sizes would settle it.
- The three morphotypes are defined from clusters in a t-test matrix rather than from an explicit statistical clustering model; a mixture-model fit to the turning-angle distributions would show whether three is the natural number of components or an artefact of thresholds.
- If the dominant morphotype is truly stable across five regions within a narrow time window, the same analysis applied to other Cruziana ichnospecies should reveal similarly conserved morphotype structure, making behaviour a phylogenetically informative trait.
- The paper's inferred time axis could be calibrated, not just assumed, by using speed estimates from living analogues; that would turn qualitative statements about 'temporary resting' into quantitative durations.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes a quantitative methodology for discretizing fossil movement paths, converting trace-fossil trajectories into turning-angle distributions, and then applying Welch's t-tests at specimen, subgroup, and locality levels. The method is demonstrated on Cruziana semiplicata from Spain, Oman, Poland, Russia, and Wales. Three previous assertions about C. semiplicata behaviour are tested: whether resting (Rusophycus) changes subsequent movement, whether Spanish subgroups show one stereotyped behaviour, and whether geographic groups differ. The authors report that resting does not alter behaviour, that locality-level differences largely disappear after removing specimens identified as 'different' by pairwise t-tests, and that three morphotypes (one dominant, two subordinate) persist across localities. The central biological conclusion is that C. semiplicata tracemakers show a dominant stereotyped behaviour with two recurring variants, and that Seilacher's two-ichnosubspecies split is not supported.
Significance. If the central morphotype claim were established, this would be a useful contribution: it imports movement-ecology metrics into ichnology and offers an open, repeatable pipeline for quantifying trace-fossil trajectories. The digitization, width-normalized segmentation, and turning-angle calculation are clearly described and represent a genuine methodological advance. The paper also makes concrete, testable claims about behaviour persistence and morphotype structure. However, the statistical evidence for the three morphotypes and their cross-locality persistence is not currently convincing because the groups are defined post hoc from the same pairwise t-test matrix used to test locality effects, and non-rejection of the null hypothesis is repeatedly treated as evidence of similarity without power analysis or multiple-comparison control. The methodological toolkit itself is promising, but the demonstration of its utility rests on inferences that need stronger statistical support.
major comments (4)
- [§3.7.4, §3.8.4, §3.9] The 'different' specimens are defined using the same specimen-vs-specimen t-test matrix that is then used to reinterpret group differences. Section 3.7.4 sets a p<0.1 threshold against ≥25% of specimens, §3.8.4 removes those 40 specimens and re-runs the group and subgroup t-tests, and §3.9 uses the black blocks of the same matrix to define morphotypes 2 and 3. This is circular: the selection rule and the final morphotype grouping are both derived from the same pairwise test results, so the disappearance of locality differences after outlier removal is partly built into the procedure. An independent classification (e.g., clustering on the turning-angle distributions themselves, or a pre-registered outlier rule not derived from the full pairwise matrix) is needed before the persistence claim can be evaluated.
- [§3.5, §3.8.1, §3.8.3] Non-rejection of H0 is repeatedly treated as positive evidence of similarity or persistence. For example, §3.8.1 says nine pairs with p>0.1 'suggest on average a similarity', and §3.8.3 reports 'no evidence to reject H0 between the Oman and Spain L groups' as support for cross-locality similarity. This reasoning is invalid without a power analysis or an equivalence test: with Wales N=5, Poland N=3, and Russia variance (σ, as reported in §3.8.3) of 15.20 against 5.84–8.75 for other groups, a p>0.1 result has very low power to distinguish 'same behaviour' from 'inadequate sample'. The load-bearing conclusion that morphotype 1 persists across localities is therefore not established even if the morphotype description is accurate.
- [§3.9, Figure 3.10] The existence of three morphotypes is inferred from visual inspection of the p-value matrix, not from an independent quantitative cluster analysis. Working from the black (p>0.1) blocks of the same matrix that was used for outlier removal, the authors group specimens into morphotypes and then use those groups to explain locality differences. This post hoc procedure does not validate the number of morphotypes or their distinctness. The paper should provide an independent clustering or model-selection step (for example, on the full direction-adjusted turning-angle data) and quantify support for three groups versus one or two groups, with appropriate uncertainty.
- [§3.7.4, Table 3.3] No multiple-comparison correction is applied anywhere, despite thousands of pairwise Welch's t-tests. The specimen matrix alone involves 136 specimens (Table 3.3) and thus on the order of 9,000 pairwise comparisons, of which many are non-independent because the same specimens appear in multiple tests. Using an uncorrected p<0.1 threshold across this many comparisons guarantees a substantial number of false 'different' classifications, which directly affects the identification of the 40 'different' specimens and hence the morphotype grouping. Either a correction (e.g., FDR) or an explicit justification for not using one is needed.
minor comments (5)
- [§3.8.3] The text reports 'Russia had a much higher variance (σ = 15.20)' while other groups have variances between 5.84 and 8.75. σ is standard notation for standard deviation, not variance. Please clarify whether 15.20 is a variance or standard deviation, and correct the notation throughout.
- [References] In §3.5 and Figure 3.12 the source for Wales is cited as 'Crimes 1968; 1970, Plate 5', but the reference list contains only Crimes (1970). Please add the 1968 reference or correct the citation.
- [Figure 3.10 caption] The caption ends with 'while = strong evidence (p < 0.01)'; this should read 'white = strong evidence'.
- [Throughout] The paper states it 'provides an open-source groundwork', but no repository, code link, or data availability statement is included. Since the methodology is the main contribution, making the MATLAB/R scripts and the digitized coordinate data available would strengthen reproducibility.
- [§3.6.2, §3.6.3] The choice of segment-distance multiplier (0.5) is stated but not motivated or tested for sensitivity. At minimum, report whether the main conclusions change for other multipliers (e.g., 0.25, 1.0), since autocorrelation and turning-angle distributions depend on this choice.
Circularity Check
Morphotype persistence is partly self-definitional: clusters are read from the p-value matrix and same-group similarity is restated as a finding; 'different' specimens are excluded using the same cross-locality comparisons later claimed to vanish.
-
self definitional
[Section 3.9 Discussion, Figure 3.10 description (p. 75)]
"Working from the regions of no evidence to disprove H0 (i.e. black squares in Supplementary Information, Figure 3.13), we were able to further cluster the “different” specimens into two groups. When the p-value matrix is organized according to these groups (Figure 3.10), two overarching trends become clear: (1) that specimens predominantly show no evidence to disprove H0 (p > 0.1) when compared with specimens of the same group and (2) that specimens predominately show weak or stronger evidence to disprove H0 (p < 0.1) when compared with specimens of a different group."
The morphotype clusters are constructed directly from the black (p>0.1) squares of the specimen-vs-specimen p-value matrix. Therefore the 'trend' that same-group specimens show p>0.1 is a restatement of the clustering criterion, not an independent discovery. The claim that there are exactly three distinct behavioural morphotypes is an artifact of manually grouping the same matrix, so the matrix cannot serve as independent evidence for the morphotypes' existence or distinctness.
-
fitted input called prediction
[Sections 3.7.4 and 3.8.4 (pp. 70, 74)]
"To determine if these specimens were having a significant impact on the results of our subgroup-vs-subgroup and group-vs-group two-sample t-tests we re-ran these series of t-tests, this time excluding the turning angle data from the “different” specimens ... This analysis showed there was no longer evidence to reject H0 between the four Spanish subgroups (p > 0.1), nor between the Spain L and R groups (p = 0.1074), nor for the Wales and Spain L groups (p = 0.1044)."
The 'different' specimens are labelled using the full specimen-vs-specimen matrix, which includes cross-locality comparisons, and the p<0.1/25% threshold is fitted to that matrix ('adequately encompassed the remaining specimens'). Removing those specimens before re-running the locality tests preferentially removes the very specimens that drove any locality difference. The subsequent non-significance—and hence the claim that morphotype 1 persists across localities—is therefore partly built into the selection rule rather than independently demonstrated.
full rationale
The core methodological contribution—discretizing fossil paths, computing turning angles, and comparing distributions with t-tests—is self-contained and not circular. No load-bearing self-citations or imported uniqueness theorems are used. However, the central empirical claim that three behaviourally distinct morphotypes 'persisted across multiple geographic localities' is weakened by circularity in the classification and outlier-removal steps. The three morphotypes are clustered by eye from the black (p>0.1) blocks of the same p-value matrix, so the observation that specimens within a morphotype are similar is a restatement of the grouping rule. Moreover, 'different' specimens are identified by the same cross-locality pairwise comparisons, then excluded before group comparisons; the resulting disappearance of locality differences is partly a consequence of that exclusion. A separate statistical concern—treating p>0.1 as positive evidence of similarity despite very small samples (Wales N=5, Poland N=3) and no power analysis—further undercuts the persistence claim, but that is a power/interpretation issue rather than circularity. Overall, the circularity is moderate: the method is independent, but the flagship behavioural result is partially self-definitional.
Assumptions & free parameters
free parameters (4)
- segment_distance_multiplier =
0.5
- different_specimen_p_threshold =
0.1
- different_specimen_percentage_threshold =
25%
- turning_direction_adjustment_rule =
sign flip to make all specimen means negative
assumptions (4)
- domain assumption Turning angles can be treated as linear data suitable for means and Welch's t-tests.
- domain assumption Tracemaker speed is roughly constant, so equal segment distances correspond to equal time intervals.
- domain assumption p > 0.1 in a Welch t-test is evidence that two distributions are similar.
- domain assumption Visual/manual tracing of published photographs preserves path geometry well enough for quantitative comparison.
Cite this review
Pith. "Pith review of Quantifying movement: Expanding the Ichnologist toolkit." pith.science (2026). https://pith.science/paper/DNFFBECR
@misc{pith2026250901102,
author = {Pith},
title = {Pith review of: Quantifying movement: Expanding the Ichnologist toolkit},
year = {2026},
howpublished = {\url{https://pith.science/paper/DNFFBECR}},
note = {Machine review of arXiv:2509.01102}
}
read the original abstract
The trace-fossil record serves as a rich dataset to examine fossil behaviour, ecologic interactions at community level, and evolutionary trends in behaviour across geological time. Behavioural adaptations are often invoked in a variety of evolutionary hypotheses; however, few methods to quantitatively compare fossil behaviour exist. Movement paths, such as trails and trackways, are well-studied in extant-organism research where they are discretized and mathematically analyzed for behavioural strategies and trends. Here, we reference modern movement ecology research and present a methodology to discretize horizontal movement paths in the fossil record. We then demonstrate the utility of this methodology and the spatiotemporal data it collects via an analysis of the trilobite trace fossil Cruziana semiplicata and assess our results in light of three previous assertions about its recorded behaviour. Our analysis reveals the presence of three morphotypes, interpreted as three distinct behavioural variations, which persisted across multiple geographic localities and are interpreted to reflect changes in external conditions, internal states, or a combination of the two. Our research highlights the immense potential of this methodology to test behavioural hypotheses and provides an open-source groundwork for future research.
Figures
Reference graph
Works this paper leans on
-
[1]
Laing1,2, Zoe Vestrum3, Luke C
55 QUANTIFYING MOVEMENT: EXPANDING THE ICHNOLOGIST TOOLKIT Brittany A. Laing1,2, Zoe Vestrum3, Luke C. Strotz4, M. Gabriela Mángano1, Luis A. Buatois1, Glenn A. Brock2, and Lyndon Koens5,6 3.1 ABSTRACT The trace-fossil record serves as a rich dataset to examine fossil behaviour, ecologic interactions at community level, and evolutionary trends in behaviou...
work page 1953
-
[3]
The probability distribution (i.e. PDF’s) of turning angles provides a measure of the morphology of a movement path. These distributions can be used to detect the presence of movement patterns (Bartumeus et al., 2008; Long & Nelson, 2013). The means (or modes) of these distributions likewise reflect the morphology of the path, with means closer to 0 refle...
work page 2008
-
[4]
to 3 mm wide), paths are compared relative to coordinates spaced x times the specimen’s width apart (i.e. the “segment distance multiplier” in our methodology). An additional strength of this unitless approach is its resilience to fluctuations in trail width within populations as well as possible errors in scale. The segment distance multiplier can be set...
work page 2010
-
[5]
used in the analysis. All scale bars are 1 cm. 87 Supplementary Figure 3.13. Specimen vs. specimen two-sample t-test results for all C. semiplicata specimens, performed on direction-adjusted turning angle data (seg. mutl. = 0.5). Yellow lines delineate different thresholds. 88 Table 3.2. Subgroup summary data. SI= straightness index. Subgroup Avg. trail w...
-
[7]
are analyzed and grouped by locality. This suggests the original results of our subgroup-vs-subgroup and group-vs-group two-sample t-tests were affected by a disproportionate sampling of “different” movement patterns (i.e. morphotypes 2 and 3). When re-analyzed through the perspective of the three morphotypes revealed by the specimen-vs-specimen t-tests, ...
work page 2007
-
[12]
(B) Poland (Radwański and Roniewicz, 1972; Figure
work page 1972
-
[13]
(C) Russia (Jensen et al., 2011: Figure 2 & Figure
work page 2011
-
[14]
and (D) Wales (Crimes, 1968; 1970, Plate
work page 1968
Show all 15 references
-
[40]
circular scribbling
to represent a potential “circular scribbling” behaviour, inferred as a stereotyped behaviour to forage nutrient-rich regions more efficiently, employed by all four tracemakers. To examine the difference between the turning angle datasets of the four sets of paths (i.e. subgro...
2007
-
[119]
F., Lewis, M
https://doi.org/10.37570/bgsd-1982-31-08 82 Fagan, W. F., Lewis, M. A., Auger-Méthé, M., Avgar, T., Benhamou, S., Breed, G., Ladage, L., Schlägel, U. E., Tang, W. W., Papastamatiou, Y. P., Forester, J., & Mueller, T. (2013). Spatial memory and animal movement. Ecology Letters,...
1982
-
[229]
Viswanathan, G
https://www.jstor.org/stable/1305990 Bartumeus, F., Catalan, J. Viswanathan, G. M., Raposo, E. P., & da Luz, M. G. E. (2008). The influence of turning angles on the success of non-oriented animal searches. Journal of Theoretical Biology, 252, 43–55. https://doi.org/10.1016/j.j...
2008
-
[447]
Raup, David, M., & Seilacher, A. (1969). Fossil Foraging Behaviour: Computer Simulation. Science, 166(), 994–995. https://doi.org/10.1126/science.166.3908.994 Savrda, C. E. (2016). Composite ichnofabrics: Categorization based on number of ichnocoenoses and their temporal incon...
1969
-
[1947]
Biometrika, 34(1–2), 28–35
The Generalization of ‘Student’s’ Problem when Several Different Population Variances are Involved. Biometrika, 34(1–2), 28–35. https://doi.org/10.1093/biomet/34.1-2.28 86 3.14 SUPPLEMENTARY INFORMATION Figure 3.11. Example of the segmentation process of discretized fossil mov...
1997 doi
-
[2008]
Constraints of fossil datasets
was present, indicated by the presence of resting traces (e.g. Rusophycus). If travel direction was available, this was indicated via an arrow. For photographs containing multiple specimens, individual paths were indicated and numbered on a copy of the original photo for cross...
2007
-
[2010]
or via reference to modern tracemakers (Miguez-Salas et al., 2022). Quantitative studies using trace fossil data are limited and either rely on the presence of easily identifiable morphologic features such as limbs and turns (Hofmann & Patel, 1989; Hofmann, 1990; Fan et al., 2...
2022
Reviewed August 5, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.