Favoring recruitment as a conservation strategy to improve the resilience of long-lived reptile populations: insights from a population viability analysis

In long-lived species, although adult survival typically has the highest elasticity, temporal variations in less canalized demographic parameters are the main drivers of population dynamics. Targeting recruitment rates may thus be the most effective strategy to manage these species. We analyzed 1136 capture–recapture histories collected over 9 years in an isolated population of the critically endangered Lesser Antillean iguana, using a robust-design Pradel model to estimate adult survival and recruitment rates. From an adult population size estimated at 928 in 2013, we found a yearly decline of 4% over the 8-year period. As expected under the canalization hypothesis for a long-lived species, adult survival was high and constant, with little possibility for improvement, whereas the recruitment rate varied over time and likely drove the observed population decline. We then used a prospective perturbation analysis to explore whether managing the species’ immature cohorts would at least slow the population decline. The prospective perturbation analysis suggested that a significant and sustained conservation effort would be needed to achieve a recruitment rate high enough to slow the population decline. We posit that the high recruitment rate achieved in 2014 – likely due to the maintenance in 2012 of the main nesting sites used by this population – would be sufficient to slow this population’s decline if it was sustained each year. Based on the results of diverse pilot studies we conducted, we identified the most likely threats targeting the eggs and immature cohorts, stressing the need to improve reproductive success and survival of immature iguanas. The threats we identified are also involved in the decline of several reptile species, and species from other taxa such as ground-nesting birds. These findings on a little-studied taxon provide further evidence that focusing on the immature life stages of long-lived species can be key to their conservation.

Methods
Between 2012 and 2020, we conducted a yearly mark–recapture survey at the end of the males’ reproductive peak of activity (i.e. 9 primary sessions). Each primary session consisted of five secondary sessions (i.e. 5 consecutive days of sampling for 7.5 hours per day). We thus sampled the population over 45 secondary sessions. The capture teams consisted of 10 to 12 people, at least half of whom were proficient at spotting and capturing iguanas. Field assistants were divided into two-person teams, with each inexperienced person partnered with someone experienced. To ensure comprehensive and uniform coverage of the islet, we divided it into five zones that could be covered in a similar duration. Consequently, zones with dense cover or high iguana density were smaller than open zones or with few iguanas. We minimized observer bias by randomly assigning a two-person team to a zone at the start of the five-day sample period. Each two-person team systematically searched for iguanas in their zone before moving to the next zone on the following day. Therefore, all teams had the opportunity to search each zone for a total of 7.5 hours. We captured iguanas by hand or with a noose and scanned them for an existing PIT tag (11.5 x 2.12 mm; 0.1g, Trovan®, Ltd., EID, U.S.); if they lacked a tag, we inserted a new one on the ventral side of the right thigh.

We analyzed the capture–recapture data using a robust design Pradel model, which is based on two nested temporal scales of sampling: the year and the repeated intra-year sessions (Pradel, 1996). Robust design schemes allow the population to be open between the primary sessions, but assume closure for the secondary sessions (Kendall, Nichols, & Hines, 1997). Recruitment is defined as the number of adults that recruit into the population in a given year for any adult present in the population the year before (Pradel, 1996). We analyzed only adult data because juvenile Lesser Antillean iguanas (SVL < 18 cm) usually have substantially lower capture and recapture probabilities than adults. They represented only 0 to 0.7% of total captures per year and thus were too few to model age-specific parameters. Their inclusion could have generated strong recapture heterogeneity, leading to biased estimates of demographic parameters (Arsovski et al., 2018). We compared models using Akaike Information Criteria with correction for small sample size (AICc) (Hurvich & Tsai, 1989). All analyses were conducted using the MARK program (White & Burnham, 1999). (See Material and Methods - Statistical Analysis for more information).

对于长寿物种而言，尽管成体存活率通常具有最高的弹性，但渠化程度较低的种群参数的时间变异才是种群动态的主要驱动因素。因此，针对繁殖补充率开展管理或许是管控这类物种的最优策略。我们针对极危物种小安的列斯鬣蜥的孤立种群，分析了9年间收集的1136份捕获-重捕记录，采用稳健设计Pradel模型估算成体存活率与繁殖补充率。基于2013年估算得到的928只成体种群规模，我们发现该种群在后续8年间以每年4%的速率持续下降。正如长寿物种渠化假说（canalization hypothesis）所预期的那样，成体存活率较高且保持稳定，几乎不存在提升空间；而繁殖补充率则随时间发生波动，且很可能是导致观测到的种群下降的驱动因素。
随后我们采用前瞻性扰动分析，探究管控该物种的未成熟种群组是否至少能够延缓种群下降趋势。前瞻性扰动分析结果表明，若要将繁殖补充率提升至足以延缓种群下降的水平，需要开展长期且高强度的保护工作。我们提出假设：2014年达到的高繁殖补充率——其可能得益于该种群在2012年得到保护的主要筑巢地——若能每年持续维持，便足以延缓该种群的下降趋势。基于我们开展的多项试点研究结果，我们明确了针对卵和未成熟种群组的主要潜在威胁，强调需提升鬣蜥幼体的繁殖成功率与存活率。我们所识别的这些威胁，同样也是多个爬行动物类群以及地面筑巢鸟类等其他类群物种种群下降的相关因素。这项针对研究较少类群的研究结果进一步证实，关注长寿物种的未成熟生命阶段，是实现其有效保护的关键所在。
方法
2012年至2020年间，我们于雄性繁殖活动高峰期结束后每年开展标记重捕调查，共设置9个主采样周期。每个主采样周期包含5个次采样周期，即连续5天、每天采样7.5小时。我们累计开展了45个次采样周期的种群采样。采样团队规模为10至12人，其中至少半数人员精通鬣蜥的发现与捕捉技巧。野外助手被分为双人小组，每一位无经验的助手均与一名经验丰富的人员搭档。为确保对该小岛实现全面且均匀的覆盖，我们将其划分为5个可在相近时长内完成采样的区域。因此，植被茂密或鬣蜥密度较高的区域面积，小于开阔区域或鬣蜥数量较少的区域。我们在5天采样周期开始时，通过随机分配双人小组负责对应区域的方式，尽可能降低观察者偏差。每个双人小组会系统性地在负责区域内搜寻鬣蜥，次日再前往下一个区域开展采样。如此一来，所有小组均有机会对每个区域累计开展7.5小时的搜寻工作。我们通过徒手或套索的方式捕捉鬣蜥，随后扫描其体内是否已植入被动集成应答器（PIT）标签（规格为11.5×2.12mm，重量0.1g，产自美国Trovan®有限公司EID部门）；若未植入标签，则在其右大腿腹侧植入新的标签。
我们采用稳健设计Pradel模型对捕获-重捕数据进行分析，该模型基于两级嵌套的采样时间尺度：年度尺度与年度内重复采样周期（Pradel，1996）。稳健设计方案允许种群在主采样周期之间处于开放状态，但假设次采样周期内种群是闭合的（Kendall、Nichols与Hines，1997）。繁殖补充率被定义为：某一年份内，相较于上一年存在于种群中的成体个体，新加入种群的成体数量（Pradel，1996）。我们仅分析了成体数据，因为幼年小安的列斯鬣蜥（吻肛长<18cm）的捕获率与重捕率通常远低于成体。幼年个体仅占年度总捕获量的0%至0.7%，样本量过少，无法针对年龄特异性参数开展建模。若将其纳入分析，可能会引发严重的重捕异质性问题，导致种群参数的估算结果出现偏差（Arsovski等，2018）。我们采用针对小样本量校正的赤池信息准则（AICc）进行模型比较（Hurvich与Tsai，1989）。所有分析均通过MARK软件完成（White与Burnham，1999）。更多细节详见《材料与方法——统计分析》部分。