319ReportsEcology,87(2), 2006, pp. 319–325q2006 by the Ecological Society of AmericaDENSITY DEPENDENCE IN MARINE FISH POPULATIONS REVEALEDAT SMALL AND LARGE SPATIAL SCALESDARRENW. JOHNSON1Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, California 95060 USAAbstract.Experimental manipulation of population density has frequently been usedto demonstrate demographic density dependence. However, such studies are usually smallscale and typically provide evidence of spatial (within-generation) density dependence. Itis often unclear whether small-scale, experimental tests of spatial density dependence willaccurately describe temporal (between-generation) density dependence required for pop-ulation regulation. Understanding the mechanisms generating density dependence may pro-vide a link between spatial experiments and temporal regulation of populations. In thisstudy, I manipulated the density of recently settled kelp rockfish (Sebastes atrovirens)inboth the presence and absence of predators to test for density-dependent mortality andwhether predation was the mechanism responsible. I also examined mortality of rockfishcohorts within kelp beds throughout central California to evaluate temporal (between-generation) density dependence in mortality. Experiments suggested that short-term be-havioral responses of predators and/or a shortage of prey refuges caused spatial densitydependence. Temporal density dependence in mortality was also detected at larger spatialscales for several species of rockfish. It is likely that short-term responses of predatorsgenerated both spatial and temporal density dependence in mortality. Spatial experimentsthat describe the causal mechanisms generating density dependence may therefore be valu-able in describing temporal density dependence and population regulation.Key words: density dependence; mortality; open populations; population regulation; predation;recruitment limitation; reef fish; rockfish;Sebastes.INTRODUCTIONPopulations often fluctuate in response to environ-mental variation (e.g., changes in climate, habitat, orthe abundance of other species), and only through com-pensatory changes in demographic rates can popula-tions counter such variability and persist over long pe-riods of time (Murdoch 1994). Knowledge of both themechanisms (e.g., predation, competition, disease) andconditions under which demographic rates change inresponse to population density (i.e., density depen-dence) is central to our understanding of populationdynamics (Royama 1977, Hixon et al. 2002, Turchin2003). Identifying the mechanisms driving demograph-ic density dependence, and evaluating their effects atthe scale of large populations has been difficult, es-pecially for species with complex life histories. None-theless, our ability to manage and conserve species’populations relies heavily on our understanding of pro-cesses that contribute to their regulation and persis-tence.The life history of many marine organisms, includingmost demersal (seafloor-oriented) fishes, involves a dis-Manuscript received 1 November 2004; revised 12 May 2005;accepted 17 June 2005; final version received 12 August 2005.Corresponding Editor: A. Sih.1Present address: Department of Zoology, Oregon StateUniversity, Corvallis, Oregon 97331-2914 USA.E-mail: [email protected] larval stage that is followed by juvenile andadult stages that exhibit more restricted movement inthe benthic environment. This ‘‘bipartite’’ life historyand potential for long-distance larval dispersal mayresult in the ‘‘open’’ (sensu Caswell 1978) spatial struc-ture of many marine populations, in which local adultpopulations are typically replenished by young that areproduced elsewhere and delivered by ocean currents.The addition of young to a local population (i.e., ‘‘re-cruitment’’) can be highly variable from one year tothe next, and variation in larval supply is known tocause marked fluctuations in the size and structure ofbenthic populations (reviewed by Caley et al. 1996).In the face of such variable replenishment, mechanismsthat contribute to density dependence in post-settle-ment growth, movement, and mortality are key to reg-ulating variation in population size and, eventually, re-production.Population sizes of many demersal fish species aresensitive to rates of post-settlement mortality (Shulmanand Ogden 1987, Hixon 1991, Doherty et al. 2004).Density-dependent mortality of juveniles can thereforehave large effects on the regulation of population size(Myers and Cadigan 1993, Caley et al. 1996, Hixonand Webster 2002). A growing body of experimentalstudies suggests that post-settlement mortality of reeffishes is often, but not always, density dependent (re-views by Hixon and Webster 2002, Osenberg et al.2002). In general, experimental studies of density de-Reports320DARREN W. JOHNSONEcology, Vol. 87, No. 2pendence are necessarily conducted at small scales(Harrison and Cappuccino 1995) and typically provideevidence of spatial (i.e., within-generation) density de-pendence (but see Webster 2003 for an example oftemporal density dependence in fish). It is often unclearwhether tests of spatial density dependence via small-scale experiments will be useful in describing temporal(between-generation) density dependence, which canact to regulate populations (Stewart-Oaten and Mur-doch 1990). Conversely, for some demersal fishes, ev-idence for density-dependent mortality has been in-ferred from time series observations of population den-sity (Myers and Cadigan 1993, Stenseth et al. 1999).However, in these cases no causal mechanisms havebeen established.A combination of small-scale experiments and large-scale observations over time can determine both themechanisms causing density-dependent mortality andhow these processes are likely to affect population reg-ulation. This information is especially important fordemersal fishes that are targeted by fisheries (e.g., rock-fish, cod, grouper), as evaluating density-dependent ef-fects can improve predictive fisheries models and man-agement strategies (Rose et al. 2001). In this study, Iconducted field experiments where I manipulated boththe density of recently settled kelp rockfish (Sebastesatrovirens) and the presence of predators to test thehypothesis that early post-settlement mortality is den-sity dependent and caused by predation. I also usedannual counts of rockfish within nearshore habitat (kelpbeds