Context-dependent dispersal directionality in a headwater stream salamander

Technical Summary: Context-Dependent Dispersal Directionality in a Headwater Stream Salamander

Problem Statement
Dispersal is a critical component of population dynamics involving fitness costs and benefits. While behavioral adaptations like non-random directionality are known to minimize dispersal costs, most existing research focuses on single-risk or single-benefit systems. In natural environments, organisms face multiple co-occurring risk factors that may influence directionality in conflicting ways. The authors note that distinguishing the effects of these competing factors in complex, multi-dimensional landscapes is difficult. Consequently, the degree to which observed directionality patterns result from multiple simultaneous risks remains unclear. This study addresses the gap in understanding how competing environmental risks shape dispersal directionality in a linear habitat where movement is restricted to a single axis (upstream vs. downstream).

Methodology
The study utilized 12 years of spatially explicit capture-mark-recapture (CMR) data from three independent headwater streams (Bear, Paradise, and Zigzag Brooks) within the Hubbard Brook Experimental Forest, New Hampshire. The study species was the spring salamander (Gyrinophilus porphyriticus), a lungless salamander with distinct larval and adult stages.

• Data Collection: Surveys were conducted weekly from June to September across 2012–2015 and 2018–2021. Researchers flipped cover objects at 1-meter intervals along 1,000 meters of stream per site. Individuals were marked with visual implant elastomer (2012–2015) or PIT tags (2018–2021). A total of 1,375 recaptured individuals (484 in Bear, 491 in Paradise, 400 in Zigzag) were analyzed.
• Dispersal Definition: Dispersal was defined as a net movement greater than 4 meters (the approximate home range size) from the initial capture position. Individuals were classified as non-dispersers, upstream dispersers, or downstream dispersers.
• Risk Quantification:
  • Predation/Flood Risk: Brook trout (Salvelinus fontinalis) densities were quantified using minnow traps in downstream reaches. High-flow events (floods) were noted as more common in downstream areas.
  • Desiccation Risk: Stream drying was proxied by the salamander's initial capture position (distance from the mainstem), as watershed area and flow stability decrease upstream.
  • Conspecific Density: Estimated using Jolly-Seber models applied to 25-meter spatial bins.
• Statistical Analysis: Logistic regression models were constructed to estimate the likelihood of dispersal (propensity) and directionality (upstream vs. downstream) for larvae and adults separately. Model selection was based on Akaike's Information Criterion corrected for small sample sizes (AICc).

Key Results

• Dispersal Propensity: Dispersal propensity (the likelihood of moving regardless of direction) was not significantly associated with any explanatory variables (location, density, or life stage). Approximately 31% of all recaptured individuals dispersed, with rates consistent across the three streams (32%, 29%, and 30%).
• Life Stage Differences: Larval dispersal directionality showed no strong association with environmental variables; the null model was the best fit. In contrast, adult dispersal directionality was strongly context-dependent.
• Context-Dependent Directionality: Adult salamanders exhibited a predictable reversal in dispersal direction based on their spatial location:
  • Adults in upstream reaches (high risk of desiccation) were approximately 1.68 times more likely to disperse downstream.
  • Adults in downstream reaches (high risk of predation by trout and flooding) were more likely to disperse upstream.
• Dispersal Distance: While adults moved further on average than larvae, dispersal distance was not strongly related to direction. The data showed a leptokurtic distribution, with most individuals moving short distances (median 18m) and few moving long distances (>200m).

Significance and Claims
The paper claims that context-dependent dispersal directionality is a significant source of individual-level variation in dispersal behavior for G. porphyriticus. The authors argue that the observed pattern represents an adaptive response to a "risk landscape" where individuals move away from the dominant local threat (desiccation upstream, predation/flood downstream).

The authors emphasize that failing to account for this spatial contingency can lead to incorrect conclusions. If individuals were pooled without regard to location, the data would suggest a strong overall downstream bias driven by the high abundance of salamanders in upstream reaches. However, the study demonstrates that individuals respond to local environmental variation, meaning dispersal directionality is contingent on spatial position.

The paper concludes that this phenomenon may be common but often overlooked in multi-dimensional landscapes, potentially appearing as "random" directionality when multiple risks are averaged out. The authors suggest that incorporating context-dependent directionality into movement models could improve predictions of range shifts for threatened or invasive species and refine management strategies regarding immigration and emigration rates. They note that this behavior persists even in a species with low perceptual range and limited dispersal ability, suggesting it may be a more widespread mechanism than previously assumed.