Evaluation of distance estimation by observers with point counts and distance sampling methodology: A field test

The distance sampling methodology for estimating bird abundance assumes that distances between the observer and the birds are measured without error. Therefore, accurate distance estimates are crucial in bird surveys. Since most detections are based on auditory rather than visual observations (i.e., singing males, usually the easiest individuals to detect), estimating the distance to heard but unseen individuals is challenging and prone to error in dense, forest habitats. In this study, we evaluated the accuracy of distance estimation by observers under natural conditions in a temperate forest of the Białowieża National Park, Poland, by conducting qualitative ‘distance band tests’ and quantitative ‘distance tests’. We found that probability of assigning an individual bird to the incorrect distance band (regardless of whether the distance was under- or overestimated) increased with distance. In quantitative tests, where we estimated the error in meters, both the frequency and magnitude of error also increased with distance. Underestimating distances, i.e., an individual was farther than the observer assumed, occurred more frequently than the opposite, i.e., an individual was closer than the observer’s assessment. While the magnitude and direction of errors varied between species, the consistent pattern for most species was that estimates were relatively accurate at closer distances but increasingly underestimated beyond ~100 m. Seasonal changes in vegetation did not influence error probability or size. Distances to singing individuals were underestimated at significantly shorter distances for higher-frequency singers, while song amplitude showed an insignificant trend. Our findings underscore the need to account for distance estimation errors in avian population research, particularly when using distance sampling. The results also indicate that assigning an individual to a distance band instead of estimating exact distance may reduce errors. In temperate forests, making the most distant band sufficiently wide to reduce errors in assigning individuals to bands may also improve accuracy.用于估算鸟类种群丰度的距离抽样（distance sampling）方法假设，观测者与鸟类之间的距离测量不存在误差。因此，精准的距离估算在鸟类调查中至关重要。由于绝大多数鸟类侦测依赖听觉而非视觉（即通过鸣唱的雄鸟——这类个体通常最易被观测到），在茂密的森林生境中，估算仅闻其声却未见其形的个体与观测者之间的距离极具挑战性，且极易产生误差。本研究在波兰比亚沃维耶扎国家公园（Białowieża National Park）的温带森林自然环境中，通过开展定性‘距离带测试’与定量‘距离测试’，评估了观测者的距离估算精度。研究发现，将单只鸟类划分至错误距离带的概率（无论距离被低估还是高估）随距离增加而升高。在定量测试中，我们以米为单位量化误差，误差的发生频率与误差幅值同样随距离增加而增大。低估距离的情况（即个体实际距离比观测者预估的更远）较高估距离的情况（即个体实际距离比观测者预估的更近）更为常见。尽管误差的幅值与方向因物种而异，但多数物种呈现出一致规律：在较近距离下估算结果相对准确，但当距离超过约100米时，估算值会愈发趋于低估。植被的季节变化对误差概率与误差大小均无显著影响。对于鸣唱个体而言，鸣唱频率更高的物种，其距离被低估的临界距离越短；而鸣唱振幅则未表现出显著相关趋势。本研究结果强调，在鸟类种群研究中，尤其是使用距离抽样方法时，必须考虑距离估算误差的影响。研究结果同时表明，相较于估算精确距离，将个体划分至距离带的方式或许能降低误差。在温带森林中，将最远的距离带设置得足够宽，以减少个体被划分至错误距离带的概率，同样可提升估算精度。