Mangrove forest phenology at the regional scale have been poorly investigated and its driving factors remain unclear. Multi-temporal remote sensing represents a key tool to investigate vegetation phenology, particularly in environments with limited accessibility and lack of in situ measurements. This paper presents the first characterisation of mangrove forest phenology from the Yucatan Peninsula, south east Mexico. We used 15-year time-series of four vegetation indices (EVI, NDVI, gNDVI and NDWI) derived from MODIS surface reflectance to estimate phenological parameters which were then compared with in situ climatic variables, salinity and litterfall. The Discrete Fourier Transform (DFT) was used to smooth the raw data and four phenological parameters were estimated: start of season (SOS), time of maximum greenness (Max Green), end of season (EOS) and length of season (LOS). Litterfall showed a distinct seasonal pattern with higher rates during the end of the dry season and during the wet season. Litterfall was positively correlated with temperature (r = 0.88, p <0.01) and salinity (r = 0.70, p <0.01). The results revealed that although mangroves are evergreen species the mangrove forest has clear greenness seasonality which is negatively correlated with litterfall and generally lagged behind maximum rainfall. The dates of phenological metrics varied depending on the choice of vegetation indices reflecting the sensitivity of each index to a particular aspect of vegetation growth. NDWI, an index associated to canopy water content and soil moisture had advanced dates of SOS, Max Green and EOS while gNDVI, an index primarily related to canopy chlorophyll content had delayed dates. SOS ranged between day of the year (DOY) 144 (late dry season) and DOY 220 (rainy season) while the EOS occurred between DOY 104 (mid-dry season) to DOY 160 (early rainy season). The length of the growing season ranged between 228 and 264 days. Sites receiving a greater amount of rainfall between January and March showed an advanced SOS and Max Green. This phenological characterisation is useful to understand the mangrove forest dynamics at the landscape scale and to monitor the status of mangrove. In addition the results will serve as a baseline against which to compare future changes in mangrove phenology due to natural or anthropogenic causes.

中文翻译：

---

红树林物候遥感及其环境驱动因素

区域尺度的红树林物候研究很少，其驱动因素尚不清楚。多时相遥感是研究植被物候的关键工具，特别是在可及性有限和缺乏现场测量的环境中。本文首次介绍了墨西哥东南部尤卡坦半岛的红树林物候特征。我们使用源自 MODIS 表面反射率的四个植被指数（EVI、NDVI、gNDVI 和 NDWI）的 15 年时间序列来估计物候参数，然后将这些参数与原位气候变量、盐度和凋落物进行比较。离散傅立叶变换 (DFT) 用于平滑原始数据并估计四个物候参数：季节开始 (SOS)、最大绿化时间 (Max Green)、季末 (EOS) 和季长 (LOS)。凋落物呈现出明显的季节性模式，在旱季结束时和雨季期间发生率较高。凋落物与温度（r = 0.88，p <0.01）和盐度（r = 0.70，p <0.01）呈正相关。结果表明，虽然红树林是常绿树种，但红树林具有明显的绿色季节性，与凋落物呈负相关，普遍滞后于最大降雨量。物候指标的日期因植被指数的选择而异，反映了每个指数对植被生长特定方面的敏感性。NDWI 是一个与冠层含水量和土壤水分相关的指数，其 SOS、Max Green 和 EOS 的日期提前，而 gNDVI、一个主要与冠层叶绿素含量相关的指数已推迟日期。SOS 介于一年中的第 144 天（旱季末）和 DOY 220（雨季）之间，而 EOS 发生在 DOY 104（旱季中期）到 DOY 160（雨季早期）之间。生长季节的长度介于 228 至 264 天之间。1 月至 3 月期间降雨量较大的站点显示出先进的 SOS 和 Max Green。这种物候特征有助于了解景观尺度上的红树林动态并监测红树林的状况。此外，结果将作为基准，用于比较由于自然或人为原因导致的红树林物候未来变化。SOS 介于一年中的第 144 天（旱季末）和 DOY 220（雨季）之间，而 EOS 发生在 DOY 104（旱季中期）到 DOY 160（雨季早期）之间。生长季节的长度介于 228 至 264 天之间。1 月至 3 月期间降雨量较大的站点显示出先进的 SOS 和 Max Green。这种物候特征有助于了解景观尺度上的红树林动态并监测红树林的状况。此外，结果将作为基准，用于比较由于自然或人为原因导致的红树林物候未来变化。SOS 介于一年中的第 144 天（旱季末）和 DOY 220（雨季）之间，而 EOS 发生在 DOY 104（旱季中期）到 DOY 160（雨季早期）之间。生长季节的长度介于 228 至 264 天之间。1 月至 3 月期间降雨量较大的站点显示出先进的 SOS 和 Max Green。这种物候特征有助于了解景观尺度上的红树林动态并监测红树林的状况。此外，结果将作为基准，用于比较由于自然或人为原因导致的红树林物候未来变化。1 月至 3 月期间降雨量较大的站点显示出先进的 SOS 和 Max Green。这种物候特征有助于了解景观尺度上的红树林动态并监测红树林的状况。此外，结果将作为基准，用于比较由于自然或人为原因导致的红树林物候未来变化。1 月至 3 月期间降雨量较大的站点显示出先进的 SOS 和 Max Green。这种物候特征有助于了解景观尺度上的红树林动态并监测红树林的状况。此外，结果将作为基准，用于比较由于自然或人为原因导致的红树林物候未来变化。