{"id":"ae355859-2557-4c3c-a29e-35a6ddbd8bbe","arxiv_id":"2505.01167","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Spider monkey core-range overlaps form persistence barcodes with higher-dimensional holes, interpreted as the group pooling complementary foraging knowledge that no single individual holds.","lead":"This paper converts spider monkeys' seasonal core ranges into higher-order overlap networks and reports topological holes that it interprets as evidence of complementary foraging knowledge shared across the group. It matters because it offers a quantitative, topology-based route to detecting distributed knowledge in fission-fusion animal societies.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No null model for range placement means the topological 'complementarity' could be an artifact of habitat geometry rather than information sharing.","rationale":"The reader's weakest_assumption is the core-range-equals-knowledge proxy, which is indeed the crux of the interpretive leap from overlap geometry to collective cognition. My concern is a concrete elaboration of that same point: even granting the proxy, the paper never compares the observed overlap topology to a null expectation for randomly or independently placed ranges under the same habitat and home-range constraints. The w* optimisation is a parameter-free ideal and is internally coherent, but it does not serve as a null model for the persistent homology results; the holes are only informative about complementarity if real sharing produces more or different homology than geometry alone. The paper's own evidence undercuts the strong version of the claim: the seasonal FCI difference is non-significant (p = 0.17), the fruit-variation correlation is non-significant (p = 0.16), and the nmax = 6 cutoff is presented as chosen to avoid trivial structure. These facts, together with the untested proxy, mean that the descriptive topology is solid but the central interpretive claim is not yet established. I therefore agree with the CONDITIONAL verdict and would not move it; the recommendation is to require the null-model analysis and a behavioural validation of the core-range proxy before upgrading to full acceptance.","tokens_in":27294,"tokens_out":1821,"duration_ms":17835,"concrete_test":"Run a spatially explicit null model: place the same number of core ranges per season as random or spatially clustered convex/alpha-convex polygons with the same individual area distribution and the same group-level home-range constraint (e.g., habitat mask around Punta Laguna), but with no knowledge sharing and no social attraction. Compute the same persistence barcodes, FCI, and w(n) curves for 1000 null replicates. If the null model reproduces the observed Betti-number persistence and the non-significant dry-vs-wet FCI difference (F1 = 2.2, p = 0.17), then the holes cannot be attributed to information sharing. Additionally, repeat the FCI comparison with nmax ∈ {4,5,7,8,9} to check whether the post hoc nmax = 6 choice drives the qualitative pattern.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim that persistent holes in core-range overlap complexes reveal complementary foraging information depends on a chain of interpretation that the paper does not test. Even granting the authors' proxy assumption that 60% a-LoCoH core ranges represent an individual's foraging knowledge for a season, the paper does not compare the observed simplicial structure (Betti numbers, persistence, boundedness of components) against any null model for how independently placed, geographically constrained ranges would overlap. The authors explicitly state in the Introduction that 'we assume that an individual's core range represents, for a given season, the area where it knows the location of available fruiting trees relatively well,' but they never validate this assumption with the available behavioural data (e.g., known fruiting-tree visits from ref. [11], or the phenology data). Moreover, the Methods note that the maximal simplex size nmax = 6 was 'the largest which consistently resulted in non-trivial structures across all seasons,' chosen upon observation of the data; since the filtration and FCI are computed on structures defined by this post hoc cutoff, the headline Betti-number findings and the absence of seasonal differences (F1 = 2.2, p = 0.17) are conditional on that choice. The theoretical w* = 1/(n+1) derivation is internally structured, but its applicability to the empirical w values is not tested: the observed w values are multimodal and often larger than predicted (Results, Figure 4), so the filtration index alpha depends on a scaling whose biological target is only partially supported. Thus the load-bearing weakness is not the absence of a single equation but the absence of any explicit null model separating geometry from information-processing claims.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper applies persistent homology to the partial overlaps of individual spider-monkey core ranges across six years, split into dry and wet seasons. The authors define an intersection/union ratio w for every subset of n core ranges, derive a parameter-free optimal benchmark w* = 1/(n+1) from an information-transfer model in the Supplementary Information, and use deviations from w* as a filtration parameter α. They report persistent holes (Betti numbers) in the resulting simplicial complexes, interpret them as \"pockets of unique knowledge\" reflecting complementary foraging information, and introduce a filtration complementarity index (FCI) to compare seasons. They also examine maximal simplicial degree centrality across simplex sizes and find no robust demographic correlates. Code and data are deposited in a Zenodo repository.","tokens_in":27510,"tokens_out":6019,"duration_ms":67536,"significance":"If the central inference were supported, the paper would provide one of the first empirical demonstrations that higher-order overlap geometry in a fission-fusion society encodes complementary information, linking topological data analysis to collective intelligence. The derivation of w* = 1/(n+1) is a genuine parameter-free benchmark, the Supplementary Information proof is detailed, and the descriptive pipeline is reproducible from the deposited code and data. The paper is also honest about the proxy assumption regarding core ranges and about the low statistical power of the seasonal comparison. However, the leap from \"holes exist in core-range overlap complexes\" to \"the group shares complementary foraging information\" is currently untested against any null model or behavioural validation, which is the central weakness.","major_comments":[{"comment":"The claim that persistent holes 'reveal complementarity in the foraging information' is not tested against any null model for how independently located, geographically constrained core ranges would overlap. A null model that randomly places or relabels the observed core-range polygons within the study area, or permutes spatial locations while preserving polygon shapes and habitat boundaries, would show whether the observed Betti numbers and FCI values exceed chance level. Without such a baseline, the holes could be produced by the lake boundary, the shape of the forest fragment, or sleeping-site centrality rather than by sharing of foraging information.","section":"Results: 'The structure of simplicial complexes shows holes at various dimensions'; Methods: 'Filtration procedure'"},{"comment":"The entire interpretation rests on the unvalidated assumption that a 60% a-LoCoH core range represents the area where an individual 'knows the location of available fruiting trees relatively well.' The paper cites previous work ([11], [20]) but does not validate the proxy with the available behavioural data, such as visits to fruiting trees or the phenology records described in the Methods. Because holes are interpreted as 'pockets of unique knowledge,' the central conclusion is conditional on this assumption; the authors should either test it against known fruiting-tree locations or explicitly reframe the claims as being about overlap geometry rather than cognition.","section":"Introduction, third paragraph"},{"comment":"The maximal simplex size nmax = 6 was selected post hoc as 'the largest which consistently resulted in non-trivial structures across all seasons.' The headline persistence barcodes and the FCI (Eq. 5) are computed on structures truncated at this data-dependent cutoff, and no robustness analysis is presented for Betti numbers or FCI across nmax (the Supplementary Information only checks centrality). The authors should show that the main qualitative results, including the multi-dimensional holes and the seasonal FCI comparison, are stable for nmax in, say, {4, 5, 6, 7, 8, 9}; otherwise the central results are conditional on a parameter chosen after seeing the data.","section":"Methods, 'Simplical complex construction'"},{"comment":"The benchmark w* is derived under the assumption that all n core ranges have identical area (N_k = N), but the empirical core ranges vary substantially in area within seasons (Figure 3). Because w is the ratio of intersection to union, a single small core range can drive low w values for large n, and the filtration defined by Eq. (4) will then classify such sets as 'less redundant' even under a null of random overlap. The authors should test whether the persistence results are robust to using an area-corrected overlap measure or to restricting comparisons to sets with comparable range sizes.","section":"Supplementary Information, Section 1.2; Eq. (3)"}],"minor_comments":[{"comment":"The term 'simplical' is misspelled in several places, including the Methods heading 'Simplical complex construction'; it should be 'simplicial.'","section":"Throughout"},{"comment":"The axis labels 'n.d12' and 'w.d12' are cryptic; the figure would be clearer with labels such as 'number of individuals n' and 'intersection/union ratio w' for each panel.","section":"Figure 4"},{"comment":"The ANOVA on observed w values treats each subset of individuals as an independent observation, but subsets overlap heavily (the same individual appears in many sets). This should be acknowledged, and a permutation-based or mixed-effects approach would be more appropriate.","section":"Results, 'The extent of overlap varies predictably...'"},{"comment":"The choice of f(alpha) = 5^{-alpha} is arbitrary, and while the authors correctly note that a monotone reparametrization preserves the filtration order, the FCI values and statements about persistence 'for several units of alpha' are not invariant to that choice. A brief statement that conclusions are robust to alternative decreasing functions f would strengthen the presentation.","section":"Methods, 'Filtration procedure'"},{"comment":"The sentence beginning 'Conversely, values below the predicted line are rare' is slightly ambiguous because Figure 4 shows many points both above and below the w* curve; it would help to specify that the statement refers to the bulk of the distribution or to a particular subset of seasons.","section":"Discussion"}],"recommendation":"major_revision","confidential_remarks":"The missing null model and the unvalidated proxy assumption are load-bearing but fixable with the existing data and code; if addressed, this would be a solid contribution to q-bio.QM and to the empirical TDA literature. The paper is within the journal's scope, and the authors have been commendably explicit about their assumptions and limitations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a serious paper with a real theoretical result and a genuinely new way to look at home-range overlaps, but the title's claim about complementary information sharing outruns the evidence. The w* = 1/(n+1) derivation is a nice piece of work—parameter-free, transparent assumptions, and it produces a concrete benchmark that the empirical w values actually trend toward. That alone is worth a look. Applying persistent homology to polygon overlaps is also new in this field, and the authors ship code and data, so the descriptive Betti-number results should be reproducible.\n\nThe soft spots are where the interpretation gets heavy. The biggest one is the null model problem: holes in a simplicial complex built from overlapping polygons can emerge from habitat geometry, shared sleeping sites, or simple spatial constraints. The paper never tests its observed structure against randomly placed or null ranges. Without that, 'complementary information sharing' is an assertion, not a result. The core-range-equals-knowledge assumption is stated explicitly in the Introduction (third paragraph), so it is not hidden—but it is load-bearing, and the available behavioral data (ref 11, phenology) could have been used to validate it at least partially. The authors do not do that.\n\nAlso, nmax = 6 is chosen post hoc ('Upon observation'), and the main comparisons—dry vs wet FCI and the fruit-variation correlation—are non-significant. The paper is honest about this, but it means the dry-season complementarity suggestion is not supported by its own statistics. The theoretical derivation is fine on its own, but its application to the observed multimodal w values is only qualitative; the authors acknowledge that overlaps are often more redundant than predicted.\n\nMy take: the math and the descriptive TDA stand, but the ecological interpretation needs a null model and better behavioral anchoring. I would send it to review, but I would insist on those changes before acceptance. The stress-test concern is on target.","headline":"A real theoretical benchmark and a novel TDA application in ecology, but the complementarity claim needs a null model before it carries weight.","tokens_in":28176,"tokens_out":1557,"would_cite":true,"duration_ms":16135,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper finds persistent topological holes in the overlapping core ranges of spider monkeys and interprets them as evidence that different individuals hold complementary foraging knowledge, so the group as a whole tracks its patchy…","keywords":["fission-fusion dynamics","spider monkeys","foraging information sharing","simplicial complexes","persistent homology","Betti numbers","core range overlap","collective cognition"],"falsifier":"A permutation null model that randomly rotates the observed core ranges around their centroids while preserving shape, then recomputes the simplicial complexes: if holes of dimension 1 and higher still persist in the randomized data, the holes cannot be evidence of knowledge complementarity.","tokens_in":26993,"feed_emoji":"🐒","tokens_out":7415,"duration_ms":67374,"temperature":0.7,"pith_summary":"The paper argues that fission-fusion spider monkeys pool foraging knowledge in a complementary way, so that different individuals know different parts of the forest and the group as a whole knows more than any single monkey. Its evidence is geometrical: the partial overlaps of individual core ranges, treated as higher-order spatial networks, contain topological holes that persist as the overlaps become less redundant. These holes are interpreted as pockets of unique knowledge held by subsets of individuals, surrounded by shared areas where that knowledge can be exchanged. If this interpretation holds, the group's constantly splitting and merging social structure is a form of collective information processing that lets it track a patchy, seasonal food supply.","feed_headline":"Topological holes in monkey ranges reveal shared foraging knowledge","feed_subtitle":"A six-year study of spider monkey core ranges finds persistent holes that point to complementary knowledge sharing.","key_machinery":"The machinery is a filtered simplicial complex: a higher-order network in which a simplex joins any set of individuals whose core ranges intersect, with each simplex weighted by its intersection/union ratio w and filtered by a parameter alpha measuring how far w sits from the predicted optimum w* = 1/(n+1). Persistent homology computes the Betti numbers, which count holes in each dimension, and their persistence across the filtration; maximal simplicial degree centrality identifies which subsets participate in the most connections.","core_discovery":"The paper's discovery is that the overlap geometry of spider monkey core ranges is balanced between redundancy and uniqueness in a way that maximizes information transfer. For any set of n individuals, an optimization model with uniform, independent movement predicts that the intersection/union ratio of their core ranges should be w* = 1/(n+1); the observed ratios follow this decreasing curve, especially for small sets. Filtering the resulting simplicial complexes from more to less redundant overlaps reveals persistent holes of dimensions 0 through 4, meaning that even as low-redundancy subsets are added, some areas are used by a subset of individuals and not by others. The authors read these holes as complementarity: shared areas provide the meeting ground where uniquely known foraging spots can be transmitted, and the persistence of the holes shows that this complementary knowledge structure is a stable feature across seasons.","pith_inferences":["An alternative reading of the holes, based on shared sleeping sites or social attraction rather than knowledge, could be tested by comparing simplicial complexes built from daytime foraging locations against those from nighttime sleeping locations; only the foraging-based complexes should show the complementarity signal.","The optimality argument assumes uniform and independent movement inside core ranges; relaxing that to movement along known travel routes would shift w*, so the observed fit to 1/(n+1) is a baseline that should be rechecked against more detailed movement data.","If the interpretation is right, the same hole-detection pipeline could be applied to other fission-fusion species, and the persistence of holes should predict the group's success at finding scarce fruit in the dry season, which the paper's small sample could not confirm.","The paper's lack of a significant dry/wet season difference, despite a visible trend, suggests the relevant environmental variable is spatial rather than temporal patchiness; testing against spatial fruit distribution is the natural next step."],"forward_implications":["If the complementarity is real, the group's collective knowledge of fruiting-tree locations is larger than any member's, and fission-fusion social dynamics are a mechanism for maintaining that distributed memory.","Subsets of intermediate size (roughly 4 to 6 individuals) should be the most important brokers of foraging information, since their overlap ratios are the most sensitive to membership changes and their simplicial centrality rises most steeply.","The persistence barcodes of the overlap complex give a season-by-season readout of how much of the group's knowledge is held uniquely, which could be monitored as the environment changes.","Because the same balance between redundant and unique area applies to any number of individuals, the w* = 1/(n+1) prediction gives a quantitative target for testing other fission-fusion species."],"supporting_citations":[{"why":"Supplies the seasonal core ranges and the overlap structure analyzed throughout the paper.","marker":"[20]"},{"why":"Prior evidence that spider monkey groups discover fruiting trees faster than individuals, motivating the information-sharing premise.","marker":"[11]"},{"why":"Agent-based model showing collective estimation of patch quality from memory and social copying, the theoretical backdrop for foraging information sharing.","marker":"[21]"},{"why":"Justifies using simplicial complexes with downward closure for higher-order spatial interactions.","marker":"[25]"},{"why":"Provides the persistent homology and Betti number framework used to detect holes in the filtration.","marker":"[32]"},{"why":"Defines maximal simplicial degree centrality used in the centrality analysis.","marker":"[31]"},{"why":"Provides the local convex hull estimator used to define each monkey's core range.","marker":"[49]"},{"why":"Computational topology software used to compute persistent homology for the complexes.","marker":"[54]"}],"fun_headline_variants":["Monkey core-range holes reveal hidden foraging information","Spider monkey range holes map shared foraging know-how","Persistent holes in spider monkey ranges encode foraging complementarity","Topological holes in monkey ranges signal complementary knowledge"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a monkey's core range is a map of what it knows about fruiting-tree locations; if the ranges instead reflect social attraction, shared sleeping sites, or habitat geometry, the holes in the overlap structure would not be evidence of complementary information sharing at all.","fun_headline_variants_meta":{"raw":{"variants":["Monkey core-range holes reveal hidden foraging information","Spider monkey range holes map shared foraging know-how","Persistent holes in spider monkey ranges encode foraging complementarity","Topological holes in monkey ranges signal complementary knowledge"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000436,"raw_usage":{"total_tokens":2175,"prompt_tokens":860,"completion_tokens":1315,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":476,"completion_tokens_details":{"reasoning_tokens":1253}},"tokens_in":476,"tokens_out":1315,"duration_ms":10056,"temperature":1.0,"reasoning_tokens":1253,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:26:22.864216+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A permutation null model that randomly rotates the observed core ranges around their centroids while preserving shape, then recomputes the simplicial complexes: if holes of dimension 1 and higher still persist in the randomized data, the holes cannot be evidence of knowledge complementarity.","supporting_citations":[{"cited_title":"E., Ramos-Fernández, G","cited_arxiv_id":null,"evidence_quote":"Supplies the seasonal core ranges and the overlap structure analyzed throughout the paper."},{"cited_title":"& Ramos-Fernandez, G","cited_arxiv_id":null,"evidence_quote":"Prior evidence that spider monkey groups discover fruiting trees faster than individuals, motivating the information-sharing premise."},{"cited_title":"& Ramos-Fernández, G","cited_arxiv_id":null,"evidence_quote":"Agent-based model showing collective estimation of patch quality from memory and social copying, the theoretical backdrop for foraging information sharing."},{"cited_title":"J., Wilber, M","cited_arxiv_id":null,"evidence_quote":"Justifies using simplicial complexes with downward closure for higher-order spatial interactions."},{"cited_title":"A., Tillmann, U., Grindrod, P","cited_arxiv_id":null,"evidence_quote":"Provides the persistent homology and Betti number framework used to detect holes in the filtration."},{"cited_title":"& Sánchez Gómez, D.Centrality measures in simplicial complexes: applications of Topological Data Analysis to Network SciencePreprint","cited_arxiv_id":null,"evidence_quote":"Defines maximal simplicial degree centrality used in the centrality analysis."},{"cited_title":"M.et al.LoCoH: Nonparameteric Kernel Methods for Constructing Home Ranges and Utilization Distributions.PLOS ONE2,1–11 (Feb","cited_arxiv_id":null,"evidence_quote":"Provides the local convex hull estimator used to define each monkey's core range."},{"cited_title":"Acknowledgements This work was partly conducted during a sabbatical stay by GRF at the Global Research Centre for Diverse Intelligences at the University of St","cited_arxiv_id":null,"evidence_quote":"Computational topology software used to compute persistent homology for the complexes."}],"review_version":1}