The importance of field sampling and data collection for scaling of ecosystem services
Authors: Brittney Roughan, David Bysouth
Introduction
When studying ecosystems, scientists often collect data in the field. For example, soil samples can be taken in the field to estimate how much carbon there is in a particular location. Often, we use local field data together with models to understand what may be happening over a larger area.
Below we show two case studies where NSERC ResNet scientists collected and processed soil carbon data in two ecosystems: salt marshes along the Bay of Fundy and forests and farmlands in the Northwest Territories.
Case Study 1: Organic carbon stocks of restored salt marshes along the upper Bay of Fundy
Step 1: Choosing representative sites and study areas
Four sites were chosen in this study to represent salt marshes at different stages of restoration: Converse - newly restored, Aulac - restored over 10 years ago, John Lusby - passively restored over 70 years ago, and Wood Point Rest Stop - natural reference site. Sites were located close to one another (<10 km apart) so environmental conditions like tides and weather were similar between sites. The basemap shows the extent of tidal marsh in the Upper Bay of Fundy along the Cumberland Basin based on geospatial data available in the Canadian National Wetlands Inventory [1].
Each site was divided into three elevation zones that are flooded by tides differently throughout the year: the low zone is flooded by ~75% of high tides in a year, the mid zone is flooded by ~50% of high tides, and the high zone is flooded by ~25% of high tides. These zones were based on flooding frequency because decomposition of organic matter by microbes is slowed down in flooded, low-oxygen conditions. The layers displayed on the basemap for the Wood Point Rest Stop site were created by simulating different tidal flooding heights over a publicly available Digital Elevation Model (DEM) for New Brunswick [24]
Step 2: Collection of sediment cores
Within a designated study area of each site, three random points within each zone were chosen for core extraction using Geographic Information System (GIS) software (ArcGIS Pro) and the location of each point was surveyed using a high-accuracy GPS system. In total, nine cores were collected from each study site. The basemap shows where the nine cores were collected in the Wood Point Rest Stop study site within the three elevation zones.
Cores were collected using a gouge auger (images below), one of the many types of ‘corers’ available to use by soil carbon scientists. Other coring types include the ‘Russian peat corer’ (extracts half a core) and manual extraction (hammering a sharpened tube into the ground). Each of these methods will create different levels of core ‘compaction’, which can affect the overall length of the core and influence the estimate of how much soil organic carbon is contained within each core. Depending on the length of the core, the process of pushing the coring device into the ground and extracting a core can be physically demanding and often requires teamwork.
Transport and handling of cores after they are taken out of the ground is another important consideration for field work. Sometimes cores are cut apart and sub-sampled in the field, and other times cores are packed in rigid plastic shells, wrapped in plastic-wrap, and refrigerated or frozen before being processed back in the lab. In salt marshes, researchers often have to walk far into study sites across uneven terrain with heavy equipment (like the corer and a cooler) to retrieve cores and bring them back.
Step 3: Processing of cores in the lab
Once back in the lab, processing the core starts. The cores are photographed and different features like soil colour and texture are described and recorded. Then cores are divided into different increments depending on the goals of the study. For this study, 5 cm increments were chosen for a total of 10 per 50 cm core, which means that 360 individual samples were processed! When dividing cores, a compression factor should be used to adjust the increment length based on the compaction measured in the field. Compaction is often assumed equal across the entirety of the core, unless compaction is measured periodically throughout the core extraction process in the field.
To calculate how much organic carbon is in a core, two measurements are needed - ‘bulk density’ and ‘organic carbon content’ or ‘% organic carbon’. Core increments are usually sampled for bulk density first, which is a measure of the dry weight of soil within a known volume of the core. In the photos below, a metal cylinder was inserted into the centre of each 5 cm increment. The soil sample was extracted using a syringe plunger and then dried in an oven (see Step 4). The same soil sample can be used to determine the organic carbon content of the increment, but others can be cut from the core and stored until they are processed.
Once processing is done, any remaining core segments are often archived and frozen for future analyses. Freezer and refrigerator space is a very important factor to consider when taking soil cores and designing the field study, as cores should be kept cold (< 4℃) to prevent decomposition of organic matter by microbes.
Step 4: Lab analyses for bulk density and loss-on-ignition (LOI)
Wet samples are placed in individually labeled tins, weighed, and then dried at 60 ℃ in a drying oven for 24-72 hours (or until a consistent dried weight is reached). After they are completely dry, they are re-weighed to determine dry weight of the soil sample. The images below show the dried soil cylinders in a desiccator, which allows the samples to cool down to room temperature without absorbing any moisture before being re-weighed. Bulk density is written as the dry weight of soil divided by the known volume of the sample (grams per cm3). Low bulk density means there is not a lot of soil within the volume of core sampled, which means the soil is less densely packed and there are more air pockets. Soils with low bulk density often have high organic matter content and water-holding capacity (like a sponge!).
After the soil samples are dried and bulk density is determined, the soil samples are often processed for loss-on-ignition (LOI), which is an estimate of how much ‘organic matter’ is in a soil sample. Before the samples are put into the furnace, the sediment needs to be crushed into a homogenous sample, often using a mortar and pestle.
LOI involves combusting the soil sample at a high temperature (550 ℃) in a muffle furnace for 4 hours. First, each sample is put in a ceramic crucible and weighed to a consistent starting weight. After the samples are combusted, they are re-weighed to determine how much mass was lost, which corresponds to how much organic matter was burnt off from the sample and is represented by ‘% organic matter’ of the sample. Sometimes accidents happen but often duplicate or triplicate samples are done for LOI, so it can take a significant amount of time to get through a set of samples. Soil carbon scientists often use different methods for LOI, which may affect the comparison of results between datasets. As there is no standard method, Heiri et al. (2001) [2] recommend reporting and being consistent in three parameters: ignition temperature, exposure time, and starting weight.
Step 5: Data processing and interpretation
Once bulk density and % organic matter are determined, the ‘organic carbon density’ of each increment of the core can be estimated. Often % organic matter (%OM) is converted to % organic carbon (%OC) in salt marshes using a known conversion equation published in the scientific literature. Some studies choose to determine the exact amount of organic carbon (%OC) in their samples by conducting ‘elemental analysis’ and creating their own %OM-to-%OC conversion equation for their specific region. This step involves more time to process samples or more money to pay for the analysis. In addition, soil samples should be acidified to remove any inorganic carbon present so elemental analysis results only reflect total organic carbon. Maxwell et al. (2023) [3] have recently published the most comprehensive conversion equation for soil organic matter (SOM) to soil organic carbon (SOC) in tidal marshes to date.
Once %OC of the increment is determined, this can be multiplied by the bulk density (g/cm3) to determine the organic carbon density of the core. Organic carbon density (g OC/cm3) is then multiplied by the height of each increment (5 cm in this example) to determine the amount of carbon within each core depth increment (g OC/cm2). Then all increments are added together to get the total ‘soil organic carbon stock’ throughout the core’s depth. This organic carbon stock (g OC/cm2 to 50 cm depth) can be scaled up to larger areas and compared to other studies. A common unit for carbon stock is megagrams (or tonnes) per hectare (Mg/ha). The figure above shows how to get from bulk density and %OC to total organic carbon stock and how to convert g/cm2 to Mg/ha using an example core from this case study.
Acknowledgements
As with any field work campaign, many hands went into collecting and processing this field data. Thank you to Elisha Richardson for providing the ArcPy script that was used to create the shapefile layers for the three flooding zones. Thank you to Emily Head, Stephanie MacIntyre, and Leila Rashid for help collecting cores in the field. Also thank you to L. Rashid for help processing bulk density and LOI samples in the lab. Further thanks to Dr. Gail Chmura for sharing accommodations with us and Dr. Jeff Ollerhead for use of his gouge auger. Many methods used in this case study to collect and process cores were based on guidance in the Blue Carbon Manual by Howard et al. (2014) [4].
Case Study 2: Soil carbon trade-offs associated with agricultural land use change in the Northwest Territories
Step 1: Choosing representative sites and study areas
We partnered with communities in the Northwest Territories, Canada such as Kakisa (shown here) to understand carbon and fertility trade-offs associated with agricultural cultivation of the boreal forest.
In 2019, a soil sampling effort was conducted to assess the soil fertility and carbon content of sites that communities were interested in for agricultural development. Study sites were selected across a gradient of potential agriculture suitability ranging from class 3 (potentially the most suitable for agriculture) to class 7 (likely unsuitable for agriculture). Sites were also selected based on input from local community members.
In 2021, a soil collection effort was undertaken to understand how soil fertility and carbon change between three different land use types: pre-cultivated forest soil, active agricultural soils, and abandoned agricultural soils. In partnership with seven farmers, 22 total sites were sampled.
Step 2: Collection of soil cores
Cores were collected using a 30cm or 60cm stainless steel soil corer to capture the entire organic layer of the site. For each transect, six soil cores were collected. Once extracted from the ground, cores were placed and sealed in plastic tubes.
Soil cores were stored in a freezer to limit decomposition processes. Upon departure from the Northwest Territories, samples were shipped back to the University of Guelph and stored at 4 degrees celsius.
Step 3: Processing of cores in the lab
The entirety of the organic layer of the soil cores was analyzed to determine soil macronutrient fertility (consisting of nitrogen, phosphorus, and potassium) as well as soil organic carbon stocks.
Soil cores were subdivided into 5cm increments to assess how variables change with depth throughout the core.
In total over 1500 lbs of soil were analyzed and processed from 2019-2022 in support of this project.
Step 4: Data interpretation
Based on the soil data we collected and the results of a global synthesis of studies, we found that soil organic carbon stocks significantly decline with increasing time since cultivation. This is concerning as the boreal forest represents the largest pool of terrestrial carbon.
We also found differing relationships between soil organic carbon and soil fertility in our study sites. In agricultural soils (both active and abandoned) soil fertility and soil carbon are positively correlated. In pre-cultivation or undisturbed sites, soil fertility and soil carbon are negatively correlated. This negative correlation is important from a land management perspective as we can target areas of high fertility, but low soil carbon to limit potential carbon losses.
We were able to use the soil data to create a map that highlights areas of lower amounts of soil organic carbon (shown in white) and high macronutrient fertility (larger circles). This serves as a preliminary land management tool that can help communities make decisions that limit carbon losses upon cultivation.
Best Practices and Opportunities
Field sampling is important for assessing many ecosystem services. However, it is important to have the following points in mind before engaging in field data collection:
- Carry out substantial background research to choose the most effective sampling methods based on the questions you are trying to answer.
- Balance the number of samples you need with the funding and resources you have to collect, process, and analyze them.
- Different studies may use different methods, even when studying the same ecosystem service, so using standard methodologies where possible allows for easier comparison of data across time and space.
- Since ecosystems often have high spatial and temporal variability, the data collected might not fully capture all the changes and differences happening over time and across the area being studied.
- Data collected for one study might not work for another study that is trying to look at ecosystem services on a different scale. Integration of modellers in field planning is important if the data is intended to be used in future studies of scaling.
- Unexpected things, like equipment breaking or bad weather, can happen during field or lab work causing missing data. Spending the time planning ensures unexpected events and missing data can be dealt with and will not influence the overall dataset.
Photos by Brittney Roughan & David Bysouth