Vegetation phenology and its interactions with climate change – a study on Turkey and its region

Abstract

In the face of climate change, one of the most crucial questions in ecology is 'How will the ecosystems respond to the impacts of climate change?'. The literature on the topic of how climate change is and will be affecting natural ecosystems is quite rich, however, it is not quite the same way around for the literature on how the natural ecosystems will react to these threats or if they will react at all. To be able to answer the initial question and prepare accordingly, first we need to understand how these natural ecosystems or vegetation cover responded to these changes before and then make healthy predictions on what will be their mechanisms to deal with the impacts of climate change in the future. Results from many studies globally already reported on the observed responses from different taxa and ecosystems to climate change. These responses can be expressed as the shifts in phenological phase timings, range and/or distribution shifts, changes in population dynamics, altered composition of communities or changes in genetic traits. Among these phenological phase timings are the most commonly reported ones. Phenology, a term coined by Belgian botanist Charles François Antoine Morren, studies the timing of periodic events in organisms' life cycles both plants (such as first leaf budding, or first leaf fall and animals (such as egg laying), investigates which factors drive these timings and in which ways they affect them and in which ways these periodic events or 'phenophases' are related to each other, how they affect each other for the same species and between different species. Plant phenology is strongly affected by climate but also capable of affecting it through, e.g., carbon sequestration, surface roughness, albedo, water, and other important biophysical cycles. Moreover, changes in phenology, even for individual species can have massive chain effects on the ecosystems, affecting many connected biotic and abiotic elements and consequently lead to changes in ecosystem functioning and processes. Phenology modeling has an important role in predicting the effects of climate change and ecosystem responses to these impacts and investigating the factors behind phenological shifts as well as in conservation and management planning. Plant phenology also became a prominent part of regional or global ecosystems simulation models and coupled biosphere/atmosphere general circulation models. As the important roles and connections of phenology to climate have been better understood, more attention and effort have started to be put into phenology studies. Phenology is traditionally tracked by field observations on individual or small groups of plants. Today it is mostly carried out by phenology networks such as USA-NPN and Observatoire des Saisons. Phenophases to be observed (such as first leaf or budburst, first leaf fall or autumnal coloring) and their descriptions are defined through the chosen protocols by individual networks. These observations produce very valuable, accurate and long-term datasets which are capable of catching critical phenophases. However, to understand the impacts of climate change on vegetation phenology and to be able to see the generalized response patterns from different vegetation covers, long-term, continuous data from broad scale areas, from ecosystem scale to hemispheric or global domains, is needed as the individual responses of plants may not represent the response of the entire ecosystem it occurs in. Moreover, observational phenology data has serious drawbacks such as not being evenly distributed around the world (e.g., some countries do not have this sort of observation records of their natural vegetation types and most available phenological records are on certain types of vegetation such as deciduous forests but rarely on ecosystems such as deserts), observations made by following different protocols, potential mistakes in records due to the observer-related errors, etc. After satellite remote sensing (SRS) entered ecological research, phenology studies gained an indisputable speed and volume as it overcomes many of the disadvantages mentioned above. SRS data provides data from local to near-global scales, datasets are continuous, long-term, and objective by the nature of the sensor measurements. With these advantages, researchers now can observe large-scale vegetation response patterns to climate change and study the long-term trends in phenological parameters such as the start of season (SOS). Phenology, measured by the SRS tools, is now named 'Land surface phenology' to emphasize the difference between them and the phenological records obtained by traditional phenology. LSP constitutes a very valuable tool to monitor phenology in the absence of observational data. Due to its location between Asia and Europe, diverse topography, the effect of the three bounding seas, three phytogeographical regions intersecting within its borders, Turkey has an enormous biodiversity richness. Turkey resides in the easternmost Mediterranean Basin which was defined as one of the most vulnerable areas to climate change impacts. Results from many studies and projections show that Turkey will be experiencing increasing temperatures and alterations in precipitation. To make accurate management, mitigation, and conservation planning for the aforementioned biodiversity richness, to have a deep understanding of if and how the natural vegetation covers of Turkey responded to climate change impacts is a must. However, Turkey is one of the mentioned countries which does not have phenological observations data for its natural vegetation covers. Phenological observations, as well as LSP studies, are made almost exclusively on plants under human care such as agricultural plants. To my knowledge there is no LSP studies, which focuses on Turkey's natural vegetation covers, investigating the long-term trend shifts in season metrics such as the start of season (SOS) at the country-scale with a clear metric extraction. To address this gap and to provide a baseline for methodology for future studies in Turkey and its region, in this thesis, I investigated the long-term (2002-2020) phenological parameter trends of two deciduous (Fagus orientalis and Quercus; Quercus robur, Quercus cerris, Quercus frainetto, Quercus pubescens) and three evergreen needleleaf species (Pinus brutia, Pinus nigra and Pinus sylvestris) by means of SRS methods and assessed their relationship with the temperature-derived parameters. I characterized the spring phenology for Fagus and Quercus in Turkey and evaluated the effect of altitude and latitude on their phenologies. As phenological parameters are very time sensitive, unlike the majority of LSP studies worldwide which utilize composite datasets, I used a daily MODIS product, 'MODIS/Terra Surface Reflectance Daily L2G Global 1 km and 500 m SIN Grid v006" (MOD09GA)' and calculated the Normalized Difference Vegetation Index (NDVI) with red and NIR band measurements for all pixels within the borders of Turkey. To keep the pixel heterogeneity minimum, I used a forest management plan to filter only the pure stands for the species I worked with and applied a vigorous temporal and spatial coverage filtering to work with pixels of the best quality for this work. This study is designed around and focuses on the parameter SOS, as the end of season (EOS) is very problematic and hard to correctly detect, and I did not have grounding observational data to be sure of my interpretations on EOS. However, to provide a full understanding on the season dynamics EOS and the length of season (LOS) parameters were also extracted and calculated and their trends over the 19-year study period was also analyzed. The evergreen species did not show phenological phase changes measurable by NDVI so they were not included in further steps of the study. I extracted the SOS parameter first, for the two deciduous species, analyzed their SOS trends over the study period and then extracted the EOS metric only for the pixels which showed a significant SOS trend (547 MODIS pixels for Fagus and 840 MODIS pixels for Quercus). LOS was calculated as the difference between SOS and EOS. Then trends in these parameters over the study period were also investigated. As temperature is considered the main driver of phenology for the temperate zone deciduous forests, I used temperature-derived variables to assess the relationship between the climate change and SOS trends of studied vegetation types. To achieve that, I calculated chilling hours (CHs), growing degree days (GDDs), mean, maximum and minimum temperature variables for different time intervals (such as FebruaryMarch or April) and with different thresholds (for CHs) and different base temperatures (for GDDs). These variable-interval-threshold/base temperature combinations were then evaluated for statistically significant changes over the study period. Variables/combinations which did not show a significant change in 19 years were not included in further analysis. According to the estimations, both Fagus and Quercus showed an earlier SOS and prolonging LOS pattern in concert with the reports from the literature. SOS was exhibiting an advancing trend over the study period by 0.8 days per year and LOS was getting longer by 1.07 days for Fagus. Similarly, for Quercus an advancing SOS pattern by 0.74 days per year and a prolonging LOS by 0.84 days per year. For Fagus, the multi-annual (2002-2020) mean SOS and EOS were found as day of year (DOY) 121.1 and 300.30 respectively and the mean LOS was 179.09 days long. For Quercus, the multi-annual (2002-2020) mean SOS and EOS were found as DOY 122.3 and 289.5 respectively and the mean LOS was 167.2 days long. The highest correlations between the SOS and temperature-derived variables were found to be February-March CH with 7.2°C threshold value for Fagus and March-April maximum temperature for Quercus.