Describe any methods used to synthesise results and provide a rationale for the choice(s). If meta-analysis was performed, describe the model(s), method(s) to identify the presence and extent of statistical heterogeneity, and software package(s) used. Elements: • If statistical synthesis methods were used, reference the software, packages and version numbers used to implement synthesis methods. • If it was not possible to conduct a meta-analysis, describe and justify the synthesis methods or summary approach used.
If meta-analysis was done, specify:
the meta-analysis model (fixed-effect, fixed-effects or random-effects) and provide rationale for the selected model.
the method used (e.g. Mantel-Haenszel, inverse-variance).
any methods used to identify or quantify statistical heterogeneity (e.g. visual inspection of results, a formal statistical test for heterogeneity, heterogeneity variance (𝜏 2 ), inconsistency (e.g. I2 ), and prediction intervals).
If a random-effects meta-analysis model was used:
specify the between-study (heterogeneity) variance estimator used (e.g. DerSimonian and Laird, restricted maximum likelihood (REML)).
specify the method used to calculate the confidence interval for the summary effect (e.g. Wald-type confidence interval, Hartung-Knapp-SidikJonkman).
consider specifying other details about the methods used, such as the method for calculating confidence limits for the heterogeneity variance. •
If a Bayesian approach to meta-analysis was used, describe the prior distributions about quantities of interest (e.g. intervention effect being analysed, amount of heterogeneity in results across studies).
• If multiple effect estimates from a study were included in a meta-analysis, describe the method(s) used to model or account for the statistical dependency (e.g. multivariate meta-analysis, multilevel models or robust variance estimation).
• If a planned synthesis was not considered possible or appropriate, report this and the reason for that decision.
SYNTHESIS METHODS (methods to explore heterogeneity) 13e Item:
Describe any methods used to explore possible causes of heterogeneity among study results (e.g. subgroup analysis, meta-regression). Elements: • If methods were used to explore possible causes of statistical heterogeneity, specify the method used (e.g. subgroup analysis, meta-regression).
• If subgroup analysis or meta-regression was performed, specify for each:
which factors were explored, levels of those factors, and which direction of effect modification was expected and why (where possible).
whether analyses were conducted using study-level variables (i.e. where each study is included in one subgroup only), within-study contrasts (i.e. where data on subsets of participants within a study are available, allowing the study to be included in more than one subgroup), or some combination of the above.
how subgroup effects were compared (e.g. statistical test for interaction for subgroup analyses).
• If other methods were used to explore heterogeneity because data were not amenable to meta-analysis of effect estimates (e.g. structuring tables to examine variation in results across studies based on subpopulation), describe the methods used, along with the factors and levels.
• If any analyses used to explore heterogeneity were not pre-specified, identify them as such.
Describe any sensitivity analyses conducted to assess robustness of the synthesised results. Elements:
• If sensitivity analyses were performed, provide details of each analysis (e.g. removal of studies at high risk of bias, use of an alternative meta-analysis model).
• If any sensitivity analyses were not pre-specified, identify them as such.
We conducted a Bayesian multinomial meta-analysis to assess how the type of habitat change (gain or loss) influences the direction of animal responses. This approach allowed us to identify whether there is a systematic pattern in how animal characteristics (such as composition, distribution, ADI, diversity, and richness) change in response to habitat alteration. A Bayesian multinomial logistic regression model was fitted using the brms package (version 2.22.0) within R (version 4.4.3).
We included unique study identifiers (auth.year) as random intercepts, To account for study-level heterogeneity, allowing for variation in baseline effects across studies. The outcome variable was the categorical direction of response, with three levels: “Decrease”, “No effect”, and “Increase”, “No effect” being the reference category for all comparisons. Fixed-effect predictors comprised the habitat change type (gain or loss), taxonomic group, and the interaction between habitat change type and taxonomic group. Additionally, we included drivers of habitat change and response variables as fixed-effect covariates to control for other potential sources of variation.
Heterogeneity
We incorporated Random intercepts for unique studu ID auth.year into the model, explicitly accounting for between-study heterogeneity in baseline response probabilities. We included an interaction term between habitat change type and taxonomic group to test whether the effect of habitat change on animal responses varies across different taxonomic groups. We assessed variation across taxonomic groups by examining the posterior distributions of these interaction terms.
Sensitivity Analyses (PRISMA 13f)
To evaluate the robustness of our synthesised results, we conducted a Prior Sensitivity test. We re-estimated the model using alternative prior distributions (specifically, normal(0, 1) and normal(0, 3) for fixed effects) to assess the influence of prior assumptions on the posterior estimates. The results obtained were consistent with the main findings, indicating robustness to prior specification. The sensitivity did not substantially alter the direction or credibility of the primary effect estimates, reinforcing the robustness of our conclusions.
Results
The analysis included 297 observations from 188 studies.The Bayesian multinomial meta-analysis revealed significant influences of habitat change type on the direction of animal responses, relative to the “No effect” category:
Habitat Loss: Habitat loss significantly increased the odds of a Decrease in animal response relative to No effect. Conversely, habitat loss significantly decreased the odds of an Increase in response relative to No effect. This indicates a substantially higher probability of decline in biodiversity indicators associated with habitat loss.
Habitat Gain: While we used habitat gain as the reference level, we interpreted its effects from the model intercepts and comparisons. The odds of an Increase in response under habitat gain (relative to No effect) were substantially higher than those observed under habitat loss. Conversely, the odds of a Decrease in response under habitat gain (relative to No effect) were relatively low and associated with higher uncertainty. Overall, these findings suggest that habitat gain significantly increases the probability of biodiversity recovery.
Interaction with Taxonomic Group
Predicted probabilities revealed heterogeneous effects across different taxonomic groups. Specifically, Bats and Birds demonstrated stronger increases under habitat gain when compared to habitat loss, suggesting a more pronounced positive response to restoration or expansion efforts. In contrast, Insects and Soundscapes showed more variable or comparatively weaker responses across habitat change types.
Heterogeneity
We observed substantial between-study heterogeneity. The standard deviation of the random intercepts (τ) for studies was estimated at 2.04 for the “Decrease” outcome (relative to “No effect”) and 2.09 for the “Increase” outcome (relative to “No effect”). This corresponds to an Intraclass Correlation Coefficient (ICC) of 0.55 (95% Credible Interval: [0.38-0.70]), indicating that approximately 55% of the total variance in the direction of animal responses was attributable to differences between studies.
