REVIEW
Abstract The amine serotonin has been suggested to
play a key role in aggression in many species of animals,
including man. Precisely how the amine functions,
however, has remained a mystery. As with other im-
portant physiological questions, with their large
uniquely identi®able neurons, invertebrate systems oer
special advantages for the study of behavior. In this ar-
ticle we illustrate that principal with a description of our
studies of the role of serotonin in aggression in a lobster
model system. Aggression is a quanti®able behavior in
crustaceans, the amine neuron systems believed to be
important in that behavior have been completely map-
ped, and key physiological properties of an important
subset of these neurons have been de®ned. These results
are summarized here, including descriptions of the
``gain-setter'' role and ``autoinhibition'' shown by these
neurons. Results of other investigations showing socially
modulated changes in amine responsiveness at particular
synaptic sites also are described. In addition, specula-
tions are oered about how important developmental
roles served by amines like serotonin, which have been
well described by other investigators, may be related to
the behaviors we are examining. These speculations draw
heavily from the organizational/activational roles pro-
posed for steroid hormones by Phoenix et al. (1959).
Key words Amine neurons á Aggression á Lobster á
Neurohormone á Serotonin
Abbreviations 5,7-DHT 5,7 dihydroxytryptamine á
5HT serotonin á A1 ®rst abdominal ganglion á
CHH crustacean hyperglycemic hormone á
CNS central nervous system á
EPSPs excitatory post synaptic potentials á
IPSPs inhibitory post-synaptic potentials á
LG lateral giant axon á MG medial giant axon á
OCT octopamine á T5 ®fth thoracic ganglion
Introduction
While observing ®ghting behavior among behaviorally
naõÈve juvenile lobsters, one cannot help but be struck by
the elegance of the unfolding scene. When placed in a
new environment, animals pause, then begin to explore
the arena, generally keeping close to the walls, which
they continually circle. Invariably they meet another
animal, and just as invariably, display their principal
J Comp Physiol A (2000) 186: 221±238
Ó Springer-Verlag 2000
E. A. Kravitz
Department of Neurobiology, Harvard Medical School,
220 Longwood Avenue, Boston, MA 02115, USA
e-mail: edward_kravitz@hms.harvard.edu
Tel.: +1-617-432-1753; Fax: +1-617-734-7557
E. A. Kravitz
Serotonin and aggression: insights gained from a lobster model system
and speculations on the role of amine neurons in a complex behavior
Accepted: 27 November 1999
weapons, the large claws. Mirroring moves, they remain
motionless, claws up, standing high on the tips of their
walking legs, or they bump, darting to and fro, fore and
aft, maintaining the display. The dactyls, the movable
®ngers of the claws, open wide but do not close to grasp
the opponent. Meetings are short, lasting about 30 s,
during which, in addition to the display, animals direct
streams of urine at each other from the nephropores at
the base of their 2nd antennae; then they break o, only
to begin exploring again. Few or many meetings may
take place, during each of which the display is repeated.
If a large size asymmetry exists, the ®ght ends usually
with the smaller animal retreating then refusing to en-
gage the larger in combat. A second component of dis-
play may be interposed with the posturing. Here is a
ballet, a pas de deux on the ocean ¯oor, in which one
animal advances, antennae whipping and claws folded
downward, while the other animal retreats, antennae
straight up and claws up and open. Then on some un-
known cue, the animals completely switch their direc-
tions of movement and the use of their appendages. If no
decision is reached with either of the displays, one of the
animals escalates to the next level of intensity, a move
immediately paralleled by the opponent. The transition
is stepwise, seamless and irreversible. Now the weapons,
the claws, start to be used, but only to grasp the oppo-
nent. Like Greco-Roman wrestlers in a giant underwater
arena, each combatant tries to overturn the other. If one
succeeds, then here too, a decision is made by the retreat
of the loser. If not, ®ghts move to the next, highest level
of intensity, a move almost always leading to a decision.
Moving with great speed now, animals advance on each
other with claws wide open. Their giant claws snap shut
on whatever they can reach, then tail ¯ips, contractions
of the large abdominal ¯exor muscles, move animals
back and upwards in attempts to tear their catch from
the opponent. The danger of damage is high now, but we
seldom see animals losing appendages. Those that do,
however, usually will not survive the continued presence
of the winner. Losers stop urinating, continually retreat,
and tail ¯ip to escape the advance of the winner.
The ``memory'' of losing persists for many days, altering
the willingness of animals to engage others in combat.
In the wild, ®ghts tend to be short in duration, with
decisions made early. Mostly these are decided by the size
dierence that exists when two random animals meet.
Under these circumstances, the making of decisions may
not have signi®cant impact on the subsequent willingness
to ®ght, but this remains to be established (references
for this paragraph: Scrivener 1971; Atema and Cobb
1980; Huber and Kravitz 1995; Karavanich and Atema
1998a, b; Breithaupt et al. 1999; Rutishauser et al. 1999).
One remarkable feature about ®ghting behavior in
lobsters is that the entire elaborate ritual, involving
stereotypical stepwise increases in intensity, and appro-
priate responses of animals to each other, all appears to
be pre-wired in the lobster nervous system. We know
this because the entire repertoire exists in socially naõÈve
animals that have been raised in complete isolation from
the 4th stage onwards. There is no indication that this
has to be learned. That is not to say that it will not be
modi®ed by experience: indeed, we already know this to
be the case (Scrivener 1971; Karavanich and Atema
1998a; Rutishauser et al. 1999). Another remarkable
feature is the long-term nature of the behavioral conse-
quences of winning and losing ®ghts. Winning animals
are more likely to win their next ®ght (Scrivener 1971),
while losers are more likely to lose again. In fact, losing
animals will not ®ght with any other animals, winners or
losers of other ®ghts, for some time after their initial
®ght. Thus, after an initial, approximately 10 min of
actual ®ghting (in an average 30-min ®ght), lobsters
appear to ``remember'' the outcome for days. Challenges
for investigators attempting to understand complex be-
haviors like aggression, are: (1) to understand how
context dependent, stereotypical, sequential pathways of
events, like the intensity-linked dierent levels of ®ghting
behavior, are assembled in the nervous system; and (2)
to ®nd out where and how these systems change to allow
experience to alter the subsequent ®ghting behavior.
As with all aspects of animal behavior, hormones and
neurohormones likely serve essential roles in modulating
aggression. We ®rmly believe that amines like serotonin
(5HT) are important in aggression in crustaceans.
However, amines are not the only, and may not even be
the most important, hormonal substances in¯uencing
aggression in these animals. In essentially all species of
animals, including man, 5HT is important in aggression
(cf. Coccaro 1989; Raleigh et al. 1991; Miczek et al.
1994; Olivier et al. 1995; Edwards and Kravitz 1997) but
evidence implicates peptides like gonadotropin-releasing
hormone and arginine vasopressin (Francis et al. 1993;
Ferris et al. 1986, 1997), and steroids like testosterone
(Rubinow and Schmidt 1996) in this behavior as well.
Recent studies have demonstrated close interactions
between amines and peptides and the neurons involved
with these substances and circulating steroids (Asmus
and Newman 1993; Bonson et al. 1994; Mani et al. 1994;
Delville et al. 1996; Ferris et al. 1997). These probably
foreshadow the ultimate unraveling of complex systems
of interdigitating humoral substances, all of which
contribute to optimal ®ghting performance. We focus
only on 5HT here, because with the exception of oc-
topamine (OCT), we have yet to positively identify other
candidate hormones important in agonistic behavior in
crustaceans. We anticipate, however, that several such
substances exist. For example, the steroid hormone
ecdysone, the lobster molting hormone, is one candi-
date, since lobster ®ghting and escape behaviors change
dramatically over the molt cycle (Tamm and Cobb 1978;
Cromarty et al. 1991). Levels of 20-hydroxyecdysone,
the active form of ecdysone, peak just before animals
show their highest levels of aggressiveness (Snyder and
Chang 1991). The peptide crustacean hyperglycemic
hormone (CHH), a putative lobster stress hormone, is a
second (Chang et al. 1999a). CHH-like peptides have
been localized very recently to a group of neurosecretory
neurons whose activity is strongly in¯uenced by both
222