ABSTRACT Accurate estimation of population density in distance sampling depends critically on the choice of detection function. Conventional models such as the half‐normal (HN) and hazard‐rate (HR) functions may perform poorly when the underlying detection structure is complex, particularly for grouped distance data. We extend the vertically reflected exponentiated hyperbolic tangent (VREHT) detection function to grouped line‐transect sampling within a likelihood‐based conventional distance sampling framework. Model performance is evaluated through an extensive Monte Carlo simulation study covering a range of detection scenarios and sample sizes. The VREHT model demonstrates greater flexibility than the HN and HR models, achieving lower bias under irregular detection scenarios while maintaining competitive estimation accuracy for regular detection patterns. These results illustrate the trade‐off between flexibility and robustness that characterizes parametric detection function models. The proposed model is further evaluated using four actual datasets representing diverse ecological conditions. Across grouping schemes, the VREHT function provides model fits and density estimates comparable to those of the standard HN and HR models while remaining robust under coarse grouping. Overall, the VREHT detection function offers a flexible and reliable alternative for analyzing grouped line‐transect data and is particularly useful when the underlying detection structure is complex.
Individual habitat patches may be connected by movement pathways to create habitat networks for resident organisms. Theory predicts that restrictions to movement among patches of these networks will impact population size and temporal stability at patch and network scales, but there have been few experimental tests of this prediction. We conducted laboratory experiments to explore the effect of reduced population connectivity on population outcomes in artificial habitat networks. We used populations of Daphnia carinata in replicate habitat networks of six increasingly complex topologies. With populations at dynamic equilibrium, all habitat networks were cut into two by removing the most central connections, disrupting population connectivity. The results showed that reduced population connectivity led to significantly reduced population sizes at both network and patch scales, but mainly in dendritic networks, most likely because of their topological differences. Linear and lattice networks maintained populations because they experienced minimal structural change and retained high connectivity post cut, respectively. At the patch scale, the magnitude of this decrease was significantly affected by the position of the patch within the network. The cut significantly reduced population temporal stability at network scales. Reduced habitat connectivity, causing population decline and destabilization, may lead to greater extinction risk, particularly for populations inhabiting less complex habitat networks with few inter-patch connections. Our study delivered a rare empirical test. Our results support existing conservation strategies that prioritize central dispersal pathways to reconnect habitats. By prioritizing population connectivity, managers can increase population size and stability, reducing extinction risk.
AIC (Akaike information criterion) is widely used for data analysis in ecology. Although AIC was designed to identify the better statistical model for prediction, this criterion is frequently used to identify the true relationship as a substitute for statistical tests. When an explanatory variable does not affect the response variable () in truth in generalized linear models (GLMs), the rate at which the model with the explanatory variable produces the minimal AIC was determined. This error rate is similar to the Type I error rate in hypothesis testing. When calculated numerically, the error rate was close to 0.157, which is the asymptotic value expected from the theory of the likelihood ratio test, irrespective of the link function and error structure in GLMs. Even very large sample sizes (100,000 or 1,000,000) do not prevent inflation of the error rate. The error rate increases with the number of explanatory variables when none of the explanatory variables affect in truth. Similar results were obtained in AICc, a modified form of AIC for small sample sizes. In conclusion, AIC or AICc should not be used as substitutes for statistical tests.
