UC Berkeley

Streams and rivers are ecosystems that are tightly coupled with adjacent ecosystems, such as forests, estuaries, and oceans. The freshwater fauna found in streams of the high Pacific Islands are all marine-derived, and many animals found in these streams have complex life cycles that involve movements between marine and freshwater ecosystems. Movements between adjacent ecosystems can provide organisms with many benefits, such as access to abundant food resources or escape from predators and competitors. The stripped mullet (Mugil cephalus) is one such organism. M. cephalus is a euryhaline fish that spawns in the marine environment, but which utilizes freshwater ecosystems as feeding habitat to varying degrees across its range. Striped mullet is a species of great concern given its important nutritional, economic, and cultural roles. In this dissertation, I delve into the ecology and behavior of M. cephalus in Hawaiian streams across large and small spatial scales.

Broad-scale factors can often be useful for understanding the distribution of stream fishes across the landscape. I start my dissertation at these broad spatial scales, where I investigated the watershed-scale factors that influence the presence-absence of M. cephalus among 33 Hawaiian watersheds. I complemented this large-scale study with a field study examining how variables at the reach-scale compared between a "mullet stream" and neighboring "non-mullet streams." For my investigation of large-scale factors, I used a model selection framework to compare 11 generalized linear models, including slope, discharge, watershed area, and watershed health. My results showed that slope was highly influential and inversely related to the occurrence of M. cephalus in lower stream reaches. In addition, discharge was also influential, but to a lesser degree, and had a positive relationship with M. cephalus occurrence. For the reach-scale comparison, I sampled depth, velocity, wetted width, and canopy cover systematically within the lower reaches of the three non-mullet streams and one neighboring mullet stream. Results from this reach-scale analysis revealed that non-mullet streams were significantly narrower and shallower, but similar to mullet streams in terms of velocity and canopy cover. This study demonstrates that slope and discharge are associated with the distribution of M. cephalus among watersheds and that, at smaller scales, mullet are associated with streams that are relatively wider and deeper.

At finer spatial scales (i.e., microhabitat), local environmental variables influence the daily decisions made by organisms, such as where to feed. Through a field study, I examined M. cephalus feeding microhabitat use within a lowland Hawaiian stream. In this section, I assessed M. cephalus feeding behavior by comparing the microhabitat variables (i.e., depth, velocity, substrate, and canopy cover) from locations where M. cephalus were observed feeding to the full suite of microhabitats available in Waiāhole Stream, O‘ahu. Data from feeding sites were collected from repeated stream bank surveys, and habitat availability data for the same study reach were collected over a two week period prior to stream bank surveys. I then used a chi-squared test to statistically compare habitat used by mullet to the habitat available. Additionally, because many of the variables are correlated with one another, I used Principle Components Analysis (PCA) to visually assess evidence of habitat selection using the suite of habitat variables. I found clear evidence that mullet are selecting specific microhabitats for feeding. The selected microhabitats were characterized by moderate depths and velocities, moderate to large substrate, and open canopies. Visual examination of PCA plots confirmed these findings. These findings suggest that M. cephalus are actively selecting their feeding locations, with individuals selecting erosional type habitats and avoiding more depositional habitats.

Finally, because I am interested in how euryhaline and/or catadromous fishes connect ocean and stream ecosystems, I studied the movement of juvenile mullet between Waiāhole Stream and the downstream estuary. Specifically, I quantified the feasibility of using Passive Integrated Transponder (PIT) tag technology to tag for studying the movement behaviors of M. cephalus. As a first step, I quantified retention for PIT tag and VIE (Visible Implant Elastomer) tags through a laboratory study. For the component, I tagged juvenile M. cephalus (7-10 cm) with PIT tags implanted in the peritoneal cavity, and VIE tags placed in post-orbital adipose tissue. Fish were kept for 48 days at the Anuenue Fisheries Research Center. I found that retention rates for PIT tags were perfect (100%) but failed completely (0%) for VIE tags. Within this information in hand, I next conducted a field study where I inferred the movement of individual M. cephalus between Waiāhole Stream and its estuary using PIT tags and a stationary RFID antenna array. For the field portion of the study, I tagged fish from Waiāhole Stream and monitored their movements between the stream and estuary using a stationary Radio-Frequency Identification (RFID) antenna array. I found that this system worked well with M. cephalus. Analysis of detections at the antenna indicated that individuals were making forays into the estuary at varying frequencies and of varying length. For example, longer forays were occurring primarily at night. Results from this work suggest that PIT tag technology is well suited for mark-recapture studies with juvenile M. cephalus, and that individuals in Waiāhole Stream made frequent forays into the estuary.

In conclusion, my dissertation work has contributed to our knowledge of the freshwater ecology of striped mullet, a globally distributed fish species that supports commercial fisheries throughout its range. Specifically, my research suggests that at large spatial scales, mullet are associated with streams that are low gradient in the vicinity of the mouth and with relatively high discharge. At smaller scales, my research suggests that mullet are feeding in erosional habitats and may be moving back and forth to the downstream estuary more regularly than has been recognized previously. Whether these patterns extend beyond the Hawaiian streams where I studied mullet is currently unknown. However, recent work suggests a high degree of genetic population structure across the global range of M. cephalus, and other work has suggested considerable life history variability across their range, including the degree of dependence on freshwater rearing habitat. In the Hawaiian streams that I studied, mullet are actively feeding and growing in freshwaters, suggesting the need for continued management efforts to conserve (and restore, where necessary) connectivity between streams and their downstream estuaries and the need for further studies to understand how these organisms use and benefit from freshwater streams and rivers across their range.