Data from: Techniques for estimating the size of low density gopher tortoise populations

The gopher tortoise (Gopherus polyphemus) is a candidate species for range-wide listing under the U.S. Endangered Species Act; hence reliable population estimates are required to aid management and guide policy needed to recover the species. Line transect distance sampling has been adopted as the preferred technique to estimate population size. However, when tortoise density is low, it can be challenging to obtain enough tortoise observations to reliably estimate the probability of detection, a vital component of the method. We suggest a modification to the method, based on counting usable tortoise burrows (which are more abundant than tortoises) and separately accounting for the proportion of burrows occupied by tortoises. The increased sample size generated by using burrows, rather than tortoises, can outweigh the additional uncertainty generated by needing to account for the proportion occupied. We demonstrate the method using surveys conducted within a 13,118 ha portion of the Gopher Tortoise Habitat Management Unit at Fort Gordon Army Installation, Georgia. An additional feature of these surveys was that we employed a systematic random design to obtain more precise estimates using a newly-developed systematic variance estimator. Individual transects consisted of pseudo-circuits, a design which greatly improved sampling efficiency on this large site. The burrow density was 0.091 ± 0.011 burrows/ha (CV = 12.6%, 95% CI = 0.071-0.116); 25% of burrows were occupied by a tortoise (CV = 14.4%), yielding a tortoise density of 0.023 ± 0.004 tortoise/ha (CV = 19.0%, 95% CI = 0.016-0.033) and corresponding population size of 297 (95% CI = 210-433). These techniques are applicable to other studies and species: surveying burrows or nests, rather than animals, can produce more reliable estimates when it leads to a significantly larger sample of detections and when the occupancy status can reliably be ascertained. Systematic line transect survey designs are practical to implement and, now, to analyze, and yield better precision.

哥法地鼠龟（Gopherus polyphemus）是美国《濒危物种法案》（U.S. Endangered Species Act）下的全域保护候选物种，因此亟需可靠的种群数量估算结果，以辅助管理工作并为该物种的恢复政策制定提供依据。样线距离抽样法（Line transect distance sampling）已被确定为估算种群规模的首选技术。然而，当龟类密度较低时，难以获取足够的龟类观测数据以可靠估算探测概率——这是该方法的核心参数。我们提出一种方法改进方案：基于可识别的龟类洞穴（其数量远多于活体龟）进行计数，并单独估算洞穴被龟占据的比例。相较于直接统计活体龟，利用洞穴作为调查对象可大幅提升样本量，这一优势足以覆盖因需估算洞穴占有率所带来的额外不确定性。

我们通过佐治亚州戈登堡陆军基地哥法地鼠龟栖息地管理单元内13118公顷区域的实地调查数据，验证了该改进方法。本次调查的另一项特色是采用系统随机抽样设计，并结合新开发的系统方差估计量以获得更精准的估算结果。每条样线采用伪回路（pseudo-circuits）设计，该方案在这片大面积调查区域中显著提升了抽样效率。

调查结果显示，洞穴密度为0.091±0.011 个/公顷（变异系数Coefficient of Variation, CV=12.6%，95%置信区间Confidence Interval, CI=0.071–0.116）；其中25%的洞穴被龟占据（CV=14.4%），由此计算得到龟类密度为0.023±0.004 只/公顷（CV=19.0%，95%CI=0.016–0.033），对应种群总数量为297只（95%CI=210–433）。

该改进技术可推广至其他类似研究与物种：当通过统计洞穴或巢穴而非活体动物能够显著提升有效观测样本量，且可可靠确定其占据状态时，该方法能够生成更可靠的种群估算结果。系统样线距离抽样设计不仅便于现场实施，且如今已拥有配套的成熟分析方法，可获得更优的估算精度。