Rillig, Ryo (joint 1st author), et al. (2019) SCIENCE  [jump to the article]

How do soil biodiversity & functions behave under many global change factors?

Global change is not only about climate change. Salinity, pesticide, nutrient cycling, plastic pollution, heavy metal contamination ... 

Almost all experimental studies related to global change focus only one or two factors together. We show that when many global change factors act together, the total effect can become unexpectedly strong. 

Anyway, see the beautiful graphical summary of this study on the left. copyright: Anika Lehmann

Ryo et al. (2019) Research Synthesis Methods  [jump to the article]

Meta-analysis can be advanced with machine learning.

Meta-analysis is a statistical method analyzing many previous studies together to find a general pattern. We show how machine learning techniques can be used for meta-analytic analysis.

Ryo et al. (2019) Trends in Ecology & Evolution  [jump to the article]

We introduce basic principles of temporal dynamics in ecology.

We figured out essential features that describe temporal dynamics by finding similarities among about 60 ecological concepts and theories. We found that considering the hierarchically nested structure of complexity in temporal patterns (i.e. hierarchical complexity) can well describe the fundamental nature of temporal dynamics by expressing which patterns are observed at each scale. Across all ecological levels, driver–response relationships can be temporally variant and dependent on both short- and long-term past conditions.

Ishiyama et al. (2019) Conservation Biology  [jump to the article]

What if removing a large dam from a river? We show that large dam removal changes both flow connectivity and regimes, which in turn, greatly changes habitat conditions downstream. 

Mansour et al. (2018) Biological Review   [jump to the article]

Why microbial community is so diverse and the community composition is complex in rivers? We applied an emerging concept -community coalescence- that describes the wholesale mixing of microbial communities for answering the question. 

Ryo et al. (2018) Journal of Biogeography   [jump to the article]

We developed a new machine learning technique to explore how important to consider nonlinearity and 3-way interactions of abiotic factors for explaining river macroinvertebrate species richness across Switzerland. At the biogeographic scale, the interactions of elevation, forest coverage, and geographic region are of critical importance.

Bergmann et al. (2017) New Phytologist   [jump to the article]

Using a large trait database (97 variables) for 141 grassland plant species, 

a machine learning discovered that the heavier the seed, the thicker the root.

That was more important than phylogeny, unexpected!

Ryo & Rillig (2017) Ecosphere   [jump to the article]

Ecology still largely relies on statistics with "linear and additive assumption." ... It this assumption always appropriate? No.

Ecology needs nonlinear and non-additive assumption!

We introduce new machine learning techniques that integrate frequentist statistics. 

Ryo et al. (2017) Ecography   [jump to the article]

Species distributions can be temporally dynamic.

Some stream fishes respond to recent flooding events based on their life history stages by changing their distributions. The short-term recent events might be more important than the long-term averaged conditions.

Ryo et al. (2016) Hydrology and Earth System Sciences   [jump to the article]

Time series variables are composed of periodic waves —  Joseph Fourier.

We developed a statistical modeling approach that decompose temperature time series data into seasonality, daily cycle, and irregular fluctuation. 

In an alpine stream, sub-daily thermal irregularity was unexpectedly large.

Ryo et al. (2015) Plos One   [jump to the article]

Flow variability drives river ecosystem dynamics. 

We proposed a scheme how to visualize and evaluate the human impacts on flow regime at a basin scale. 

Ryo et al. (2014) Journal of Hydrometeorology   [jump to the article]

Flood disaster simulation needs to be accurate. But sometimes not.

We found that the accuracy was worse "when flood magnitude is moderate and the ground is dry before  a rainfall event comes". 

Measuring soil moisture in the region is important.

Iwasaki et al. (2012) Freshwater Biology    [jump to the article]

The very first empirical study that indicates some characteristics of flow variability explains fish species richness worldwide.