In Hilario-Husain et al.1, we explored the potential link between sociopolitical conflict and biodiversity knowledge shortfalls in Mindanao in the Southern Philippines using publicly available datasets on conflict and biodiversity records. We found that species occurrence records were related to the frequency and distance of conflicts at the provincial scale. We argue that security risks, logistical challenges, and political restrictions associated with conflict zones often deter biodiversity research, leading to under-documented regions despite their ecological significance. Our study provides an initial discussion highlighting that sociopolitical and environmental conflicts represent an underestimated threat to biodiversity in the Philippines and suggests potential solutions to address them. In response to our work, Pitogo and colleagues raised some interesting concerns, specifically regarding the choice of dataset analysed and our analytical decisions. While they acknowledge the significance of our work, they suggest that it is premature and warrants reconsideration. We welcome their interest in our work and are open to their criticisms. The impressive number of authors from diverse academic institutions, including those from the Philippines, China, Taiwan, and the United States, who showed interest in our study, indicates the significance of this issue.

First, we would like to clarify that our work neither claims to investigate nor hypothesizes the direct effect of sociopolitical conflict on species richness per se (i.e., increasing conflict decreases levels of biodiversity) or on biodiversity as a whole, but our approach was exploratory, aimed at providing a basis for understanding the link between sociopolitical conflict and biodiversity knowledge shortfalls2. We frame biodiversity shortfalls by focusing on gaps in our understanding of species distribution, especially in geographic areas with limited or no data, and on knowledge accessibility gaps caused by challenges in accessing or sharing data due to conflict and war. We reiterate our overarching goals, acknowledge the preliminary nature of our study, and recognise the inherent caveats in pioneering research that lacks prior empirical groundwork1. Thus, the interpretation of our initial results requires care and caution. However, we believe that Pitogo and colleagues may have overlooked our objectives and chosen analyses, which could have led to a misunderstanding of the message we aim to convey.

Second, Pitogo and colleagues raised concerns about our data sources for analysis and stated the use of ‘vague definitions’. In Hilario-Husain et al., we used publicly available data from the United Nations Office for the Coordination of Humanitarian Affairs (UN-OCHA)3 and MOBIOS + 4 datasets, which represent conflict events and biodiversity occurrence data, respectively. Pitogo and colleagues criticised that using UN-OCHA data did not adhere to their ideal definition of ‘conflict.’ We argue that the conflict events reported by UN-OCHA were defined according to the ACLED (Armed Conflict Location and Event Data Project) Code Book5 following the internationally agreed-upon definition. Therefore, data from the UN-OCHA remain the most extensive and standardised source of information currently available and can be utilised as a proxy for sociopolitical conflicts and to remove potential bias in collating data from multiple sources.

Pitogo and colleagues also raised concerns about the potential bias in our dataset selection and the representativeness of taxonomic groups. While we acknowledge the limitations of the data coverage, we refute the claims of bias. Pitogo and colleagues recommended incorporating the entire GBIF dataset into our study, a suggestion that was also raised by an anonymous peer reviewer, but we opted to limit our data based on literature records. Our reliance on publication-based biodiversity records was intentional, as it reflects research efforts and knowledge gaps, aligning with our objective of examining changes in knowledge generation through standardised research.

The taxonomic limitations of the database have been clarified and are consistent with the available data, and our analysis was at the ‘class’ level, where errors at the species level should not affect our analysis. Similar to other biodiversity databases, some errors are inevitable, often originating from inaccuracies in species records and the reporting of occurrences in the original mobilised data sources. As with all GBIF data, these records undergo continuous verification and correction6,7,8. Regardless of these inconsistencies, each of the ‘species occurrence data’ we used in our analysis represents ‘biodiversity knowledge’, which is consistent with our intended analysis. In addition, they also raised concerns about the limited taxonomic representation in our analysis, and we emphasise that our analysis included major taxonomic groups known to be reliable environmental proxies or indicators9,10. Achieving ‘complete’ data for all groups is a common challenge in macroecological research, particularly for elusive species11,12.

Pitogo and colleagues also questioned our choice for spatial analysis. We used the simple ‘join attributes using the nearest function in QGIS, which measures the distance of the attributes of one layer (species occurrence) from another (conflict records) based on feature’s proximity. Using this function, we set the maximum nearest neighbours to 1 to ensure that only the single closest conflict record was considered for each biodiversity occurrence record. For the analysis, distance measurements for each taxonomic category were averaged across each province. It should be noted that all distances were measured in kilometres, including those presented by Hilario-Husain et al.1 in Figure 4b. To ensure clear and interpretable results, we prioritised the nearest conflict records to minimise spatial noise from distant or less relevant data.

Consequently, Pitogo and colleagues reanalysed the same dataset and found slight variations and differences compared to our outcomes; in particular, they indicated a decline in observed species occurrence records with increasing distance from conflict events, which is the opposite of our findings. These differences may have been due to the variability in our analytical choices and decisions (for example, Botvinik-Nezer et al.13, Gould et al.14, and Silberzahn et al.15). Firstly, Pitogo and colleagues, they treated taxonomic groups as a random effect rather than a fixed effect, which does not align with our primary goal and is different from our analysis. We reiterate that our aim was to estimate and compare the specific effects of each taxonomic group. We reason that if taxonomic groups are treated as random effects, the model will not estimate the individual coefficients for each group. Instead, it estimates a single variance parameter that describes variability among groups. This would obscure the specific effects of each group, which is contrary to our goal of assessing the differences in response to distance and frequency. Secondly, Pitogo and colleagues did not disclose their critique of the data treatment they made on independent variables prior to analysis (i.e., log-transformation of the distance and frequency values), which may have altered the values of our results and produced patterns that differed from ours.

We recognise the critiques regarding the conservative estimates of our models, which may stem from the uneven sampling effort in the real-world data that could have distorted the data distribution, reduced variability, and potentially masked the true strength of the observed relationship between conflict and biodiversity records. To validate and support our previous findings in Hilario-Husain et al., we filtered ‘synonymous’ and ‘doubtful’ records and reanalysed the data using a taxon-specific generalised linear mixed model with a negative binomial distribution (i.e. taking account of zero-inflated data) at the municipal scale. Our results and the observed patterns converged and remained consistent with our previous findings, but had better and refined estimate values (see Fig. 1). We suggest that the stronger effects at the municipal scale highlight the importance of a scale-dependent analysis of socioecological interactions, reflecting the mechanisms and impacts being studied. We also recognise that the provincial scale may be too coarse to detect distinct patterns that are more apparent at finer, more localised resolutions. Nonetheless, both of our analyses consistently filtered the ‘noise’ and highlighted the important ‘signal’: sociopolitical conflict could hamper biodiversity research efforts. We also incorporated additional findings showing a significant difference in the distance of occurrence records from conflict areas based on species conservation status, with under-assessed species often found closer to critical areas (Fig. 2). This could potentially undermine the effective conservation status of under-assessed or even undescribed species in conflict zones, which may face risks similar to those of well-studied groups16. However, this observation warrants further investigation to fully comprehend the underlying relationships.

Fig. 1: Municipal scale analysis of biodiversity-conflict relationship.
figure 1

Dot plot showing the estimates from the taxon-specific generalised linear mixed model (GLMM) with a negative binomial distribution (i.e. to account for zero-inflated data) explaining species occurrence records influenced by frequency (a) and distance (b) from conflict events. Each dot represents the estimates from separate GLMMs, and the whiskers represent the 95% CI. The figure was generated using GraphPad Prism 921. Note: the image icons used from https://www.flaticon.com/ under the Creative Commons Attribution-NonCommercial (BY-NC) License.

Full size image
Fig. 2: Conservation status of species in conflict areas.
figure 2

Comparison of the distance of occurrence of species records from conflict events according to their Red List conservation status. Our analysis showed a significant difference (Kruskal-Wallis test: H = 4760, df = 7, p < 0.0001) in the distance from conflict areas. The whiskers represent the 95% CI. Figure was generated using GraphPad Prism 921.

Full size image

More importantly, both our findings shed light on the critical need for careful contextualisation of our conclusions to prevent potential misinterpretation by the public. Conversely, from the opposing findings of Pitogo and colleagues, the following question arises: Why does species richness drop as you move farther from conflict zones? To make their case more compelling, Pitogo and colleagues could strengthen their argument and raise informed dialogue by offering alternative explanations to reconcile opposing findings and include alternative conflict datasets for deeper analysis. This would provide a balanced perspective and allow conservation biologists and policymakers to develop practical solutions grounded in a thorough understanding of sociopolitical conflict and biodiversity.

Sociopolitical conflicts disrupt biodiversity surveys, data collection, and conservation efforts, creating gaps in our understanding of biodiversity dynamics in the affected areas17,18. When linked to habitat loss or subsistence hunting, such conflicts can lead to biodiversity loss, which is often unmeasured because of limited data generation. Conversely, they may create areas where humans cannot access and allow wildlife populations to grow19. As awareness grows, conflicts emerge as an underestimated yet critical indirect driver of global biodiversity loss18,20. In the Philippines, the impact of warfare on biodiversity remains underexplored, and our research pioneers this field while acknowledging the need for further scientific and political priority settings to fully understand this relationship.

Finally, we reassure Pitogo and colleagues that our study is not a panacea, but only serves as an initial step to stimulate dialogue on this important yet often overlooked environmental issue. Acknowledging that our analysis does not offer a single silver bullet solution to the ongoing socioecological challenge, we welcome and anticipate further studies that will deepen our understanding and address the complex issues between the impact of sociopolitical conflict on biodiversity in the Philippines. We urge readers to consider the broader context of our study to understand the importance of our findings in addressing key ecological concerns and enhancing awareness of the negative impact of sociopolitical conflict on biodiversity. As mentioned earlier, we appreciate the criticisms from Pitogo and colleagues, as they highlight the need for further efforts in this area of environmental research. There is still much to discuss about the best way to do it. We hope that this response not only clarifies the procedures they questioned but also contributes to advancing the field and highlights the significance of our work in driving meaningful progress in biodiversity conservation efforts in Mindanao and other regions facing similar challenges due to war and conflict.