Woi Sok Oh - Climate and conflict on internal displacement: network analysis of Somali case and Jeffrey Smith - Ignoring key parameters in species distribution models undermines our ability to predict biodiversity responses to climate change: The problem

Thu, Mar 3, 2022, 12:30 pm

Woi Sok Oh - A growing number of populations are forcibly pushed out of their residence around the world. More than half of displaced populations move within a country and are called internally displaced persons (IDPs). IDPs’ movements are driven by various reasons, including conflict, drought, flood, etc. This study particularly focused on Somali IDPs who have suffered from protracted conflicts, droughts, and flooding events. Despite the severity and continuity of the problem, our knowledge is limited in how climate and conflict build IDP networks respectively. What are the underlying push and pull mechanisms in climate-induced and conflict-induced IDP networks? How are climate-induced and conflict-induced IDP networks structured and characterized? How do geographical locations cluster differently in two IDP networks? In the seminar, I will discuss the application of network analysis techniques from biology in solving these questions.

Jeffrey Smith - To prevent further losses of biodiversity due to climate change ecologists have invested significant effort developing tools to project future patterns of biodiversity. Presence-only species distribution models, and in particular MaxEnt, have become the most widely used tools for mapping the distribution of species and projecting how they will respond to climate change. However, these model predictions often fail to align with the observed responses of biodiversity to climate change.  In this paper we investigate how the commonly assumed default prevalence value in MaxEnt models or an assumed threshold of habitat suitability for species occurrence overestimate biodiversity responses to climate change, and we provide 3 solutions for generating results that are more aligned with observations. We do this by linking MaxEnt models derived from presence-only records with known presences and absences from the Breeding Bird Survey for 284 bird species. Assuming either the default uniform species prevalence in MaxEnt models or a threshold for occurrence can greatly overestimate climate change effects on species richness and range size, underestimate its effect on rare species, and misidentify of the most important areas for conservation. We show that MaxEnt models can be brought more in line with real-world observations by (1) reporting proportional rather than absolute change or change using thresholds, (2) using a uniform, biologically informed prevalence value, or (3) using globally available species attributes to approximate species-specific default prevalence values.

Location: 
Guyot 10/Virtual
Audience: 
Open to public
Speaker(s): 
Audience: 

Upcoming Events

No upcoming events found.