See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/279392263 Interactions of Seeds and their Insect Predators/Parasitoids in a Tropical Deciduous Forest CHAPTER · JANUARY 1975 DOI: 10.1007/978-1-4615-8732-3_8 CITATIONS 6 READS 9 1 AUTHOR: Daniel Janzen University of Pennsylvania 163 PUBLICATIONS 6,122 CITATIONS SEE PROFILE Available from: Daniel Janzen Retrieved on: 05 February 2016
INTERA~Jr0NS or SEEDS ·AND THEIR INSECT PREDATORS/PARASITOIDS
IN A TROPICAL DECIDUOUS ~OREST
~?niel H. Janzen
Department of' Biology, University of Michigan
-Ann Arbor, Michigan 48104, U.S.A.
In this paper I will touch on some of the patterns and processes that have come to light in an on-going study of how the insects that eat seeds in tropical vegetation influence the struct~re of that vegetation and the characteristics of its indiv~dual members, and how the characteristics of the plants influence
"the insects. This study began in 1964 in the lowlands of the s'tate of'VeraCr'llZ~ Mexico. While-trying to understand the population dynamics and interdependency of obligate acacia-ants and their
,aca'cias (Janzen 1967), I was "struck by the thoroughness with which the bruchid Acanthosce'lides oblongoguttatus decimated the seed crops.of,Acacia copnigepa. This swollen-thorn acacia was extremely c'oinrrion" as a result of farming and grazing pract ices, and it produced far more fruits than the local community of dispersal agents Hould eat (a th0roug~ study of dispersal agents was~not made, but at leas-t"-the Plain-tailed Brown Jay, Psilophinus mexicanus, BlacKheaded Saltator, Saltatop atpiceps, and Grayish Saltator, Saltatop coemlescens, ate seeds along with the pulp and defecated living seeds). l-:hile most of the A. cOPnigepa population bore flowers in the later part of the dry season (March, April) and mature fruits about ten months later, individuals in almost any reproductive stage could be found at any time of year in an area as small as a few hectares. The pods remained on the acacia in a stage susce?tible to bruchid oviposition for as long as two to three months. Attack was heavy and continuous, both from beetles that could have emerged from other pods on the same acacia, and from beetles corning from other A. copnigepa out of fruiting synchrony. Only very broad synchrony of flowering and fruiting occurred in the highly mixed and variously disturbed habitats (roadsides, back yards, various age corn fields, pastures, fencerows, creekbanks, marshes),
154
INTERACTIONS OF SEEDS AND THEIR INSECT PREDATORS/PARASITOIDS 155
It was obvious that in the highly disturbed lowlands of Veracruz, the acacia was largel" losing the race between the brucr-lids and the dispersal agents. However, it seemed likely that in a more intact habitat where A. cornigera would be much rarer, more synchronized, and further apart, the interaction would not be so onesided. It was also 9bvious that such habitats were not available to me, so t!1.e sUbj,ecl~~ she~ved while I worked on the interaction between ~he'ants and.the acaC1as.
,-
~hile working out the ecological distribution of the ants and their acacias throughout Central America during July and August of 1966 and 1967 (Janzen 1974a), I had the opportunity to make general collections of legume seeds in many habitats. Confronted with numerous:j)ags of seed,pod~,.from some of which came l~ge numbers of bl'uchids while from ,others came none, it was obvious that there was a pattern. Large-seeded legumes in general lacked bruchids while almost all of the small-seeded ones had bruchids (Janzen 196"9). It is perhaps appropriate to add here that the Janzen (1969) paper, which attempts to relate seed size, seed toxicity, and bruchid attack, contains 'two quite annoying errors. First, after developing the entire paper around the idea that species preyed on by Bruchidae have smaller seeds, more seeds, and more biomass of seeds per unit canopy volume, the first sentence of point one in the "Conclusions" section contains a ludicrous printing error, s'Jch that the sentence reads just the opposite (the word "not" belongs between the first "are" and "attacked", not the second, page 23, column l~ iines 36-40). Second, the primary biological exception to the tentative rule put forth in the paper was a very smallse'eded species of Indigofera; further collecting in Mexico and Central America has shown that this legume is heavily attacked by a. v~Py ,small species of bruchid (unpublished field notes, Johnson 1973, and Center and Johnson 1974).
Tne consideration'of large seeds as toxic was nearly as sere'ridipi tous as finding that they lacked bruchid attack. I was walking down a hall in the Botany Department at the University of Kansas holding a Costa Rican cycad fruit in myhand, and a person walking the other way commented that he had a whole refrigerator full of them and did I know that they have very toxic seeds? I said no, I did not know that, but that I had a seed upstairs that I was willing to bet would be very toxic and did he wish to analyse it. In a few hours E. Arthur Bell was back with the announcement that my Mucuna andreana seed had an extraordinarily high concentration of L-DOPA (L-dihydroxyphenylalanine), an uncommon and non-protein amino acid with well known toxic physiological effects on mammals. This led to a number of examinations of large and bruchid-free seeds for high concentrations of potentially toxic secondary compounds <e.g., Bell and Janzen 1971) and to an increasing frustration with our lack of knowledge of whether these compounds really are toxic to bruchids.
Serendipity struck again. Back in 1966 I h~d met Paul Feeny while he happened to be passing through -he University of Kansas, and he told me about his work with tannins in oa~ leaves and how he had fed them to armyworms (Prodenia eridania) in the laboratory to illustrate tannin toxicity. I.had then coll?borated witq him and Sherry Rehr to test an. o,J,d hypothesis (Janz'em 1966, 196'7) that the swollen:thorn acaciar)_~a~es would be less toxic to insects than their non-ant-acacia-- relativ-~S";-- they are less toxic (Rehr et al. 1973a) and Feep.y·was an obvious person to test the L-DOPA and other compounds thai. Bell had found in large legume seeds. L-DOPA and cther seed compounds were then·found to be toxic to armyworms, depending on concentration (Rehr et al. 1973b,c).
Locking at ~he bruchids, their interplay with seed dis pe1sal agents, and the 'to~ici ty of the-seeds to some animals but not all, suddenly caused me to realize that these and other seed predators could be not only influen9ing the genetic characteristics of individua± plant species in the habitat, but also influencing their relative abundance. By asking myself exactly what is it that decides whether an adult of tree species.X actually appears at point Y in the forest, I f01}nd myself "reinventing" all those things that I should have learned about probability theory when I "las an undergraduate. I spent about six months figuring out how to graphically represent the interact ron between the size and dispersal pattern of a seed crop, and a seed predator that is differentially effective depending on the" number of seeds present. I then asked Richard Levins about it, and he'point-ed'out that the "model" I had invented was the answer to one of the questions on the final examination in mathematical biology that he was teaching that semester. At any rate, the resulting theoretical analysis of the potential impact of seed predators on, forest community structure (Janzen',1970) forms the underlyin'g -philosophical structure for most of what follows here. Since that time, I have tried primarily to obtain a data base that would allow m~·to test some of the hypotheses presented in 1970, and have ga1:hered data so as to optimize the discovery of patterns clarifying the questions posed in the first sentence of chis paper. I will deal with a series of interrelaced but not all-en2ompassing patterns and apparent hypotheses that have come to lighc by focusing on one major vegetation type, tropical deciduous forest, in one area, the Pacific coastal lowlands of Costa Rica (primarily Guana2aste ?rovince). This particular study began in 1965 while teaching for the Organization for Tropical Studies and would not have come about without the logistic and intellectual support of tba! organization, and the students, faculty, and researchers-working with it.
DEFINITIONS
There are only two bothersome terminological snarls of concern here: seed predators and secondary compounds.
INTERACTIONS OF SEEDS AND THEIR n~SECT PREDATORS/PARASITOIDS 157
[cologically and behavionally, the adult bruchid (or weevil, cerambycid, etc.) is acting as a pred~tor (Janzen 1971a). It moves through the habitat searching out and killing individual seeds. It kills them by laying a~ egg on or near them. It is satiated by running out of eggs in a' given time interval. On the other hand, the bruchid ~rya;functions as a different sort of predator, one commonly.~allea"~p&rasitoid in the biological control literature. The lGlrva kills (tlsually') only one prey individual, does it slowly, does"not leave it, etc. From the standpoint of the adult tree, the pl~~t is losing its ,seed 'to a seed predator as surely as if it were losing them to a deer or squirrel. From the standpoint of the ecology of the insect, it is traditional to view the insect as a parasitoip .. Therefore in ~iscus~ions ~f tree ecology, fI wi~l think of them ci"s·seed predators, 5ut ~n ask~ng why there are"a g~ven number of them found on a host, it will be most profitable to think of them as parasitoids.
Secondary compounds are chemicals found in seeds (and in other plant parts) tha!: are .not part of the metabolic processes common to all or most plants. They are usually functional as floral and fruit attractant·s, antibiotics, repellants, and/or toxicants, and usually occur at much higher concentrations than do other odd compounds operating within the physiology of the plant (hormones, germination inhibitors, etc.). In this paper, my comments about secondary compounds are restricted to those that generally have a negative
. ,physiological effect on an animal (e.g. alkaloids, uncommon amino acids;'terpenes, phenols, saponins, lectins, etc.).
DENSE VERSUS DIFFUSE POPULATIONS
A core aspect of the potential impact cf pre-dispersal seed . predators on trees is that as the distance between seed crops in .space and/or time increases, the percent seed mortality within an individual tree's crop should decline. This hypothesis is extraordinarily difficult to test in nature, owing to the difficulty of interpreting the data obtained when cO>7.;-aring percent seed mortality among crops of clumped conspecifics with crops cf diffusely distributed conspecifics. To illustrate this difficulty, I here present a partial analysis of the predation on seeds of clumped and widely spaced Acacia faY'nesiana ("guisache", l1imosaceae) by two species of host-specific bruchids, Mimosestes sallaei and M. immunis. Background natural history is given before the actual test.
The study site lies along the northeastern edge of the approximately 25 km 2 forested seasonal swamp behind the hill behind the Organization for Tropical Studies (OTS) Palo Verde Field Station, which is in turn in the southwestern end of the CmlELCO ranch, which is in turn to the southwest of Bagaces, Guanacaste Province,
('-..,s~a Rica (about 100 40' N. Lat., 10 m elevi"_ion). A. far>nesiana is very .,idely distributed in C;entral and northern SOJ.lth America, and indigenous to the study site; there is no hint that either species of Mimosestes is recently intrpduced. It is impossible to knoW- how much European man, his cattle, 'and his lumbering have disturbed the site, but it appe~rs io never have been cleared for field or pasture. However, i~olated-t~~~ ha~e been cut, firing history is undoubtedly diff~vent from the conditions under which the bruchidacacia interactiori:'evolved, and the site has been foraged in by semi-feral cattle for at least -100 years' and perhaps as long as 30P years. It is about as natural a habitat for A. farnesiana as can be found in Costa Rica. The swamp and the adjacent forest still contain individuals qf what may have been dispersal agents for A. fa~esiana (white-''ta-iled deer, pee-caries, agoutis, pacas, and smaller rodents) but their density and relative proportions undoubtedly do not represent those under which the bruchid-acacia interactio,I} eyolyed. __ Cattle now occasionally eat variable amounts of A. faI'nesiana pods and likewise serve as dispersal agents, but
-'there is no way to know to what ~egree they replace the native animals as dispersal agents. '
The study site is at the edge of the swamp, lying to the north of where the swamp. edge is cut by the dirt track from the COMELCO main ranch headquarters to the Palo Verde Field Station. In those portions of._the swamp not severely damaged by roadwork, fire, and cattle" the vegetat:Lon consists of a dense stand of Parkinsonia acuZeatd ("palo verde") in the center, surrounded by concentric variable-width,~ings of PitheceZZobium duZce, CocoZoba caracasana, Acacia farnesiana, and Acacia coZZinsii. Each of these species is found abundantly mixed into the vegetation outside of the ring of
·.its greatest density. As one approaches the edge of the s''':cmp, the species richness of woody plants begins to climb very rap~dly and the adjacent hillsiqe may have as many as 150 species of trees, woody shrubs, 'a~,_ywody vines.
The dense stand of h. f'arnesiana '::"~es approximately:::':' minto the swamp and contains roughly 10 to 30 adulTs per hectare, -dith ()nly scattered other species of low trees or shrubs mixed c~.:ng them. Each A. fal>nesiana is about 1.5 to :; Tn tall, with 0 c[')wn of :> ":0 30 m circumference. It is 2 to 30 m between their C::'Clwns. ,The diffuse stand of A. farnesiana is at the edge of the s';c.-:?; the acacias occur at a density of about 1 per hectare, wiT~ SO to 100 m between conspecifics. The intervening spaces are fil,::,,~j wiTh obout 30 species of woedy shrubs and small trees and it is feDerally impossible to see one A. farnesiana when standing at another-.
In ar. earlier terminology, it would have been customary to speak of the dense stand as being the center of the population and the diffuse stand as the ed~e of the popula~ion. However, sjnce we have not the slightest idea of where the seeds corne from to
INTERACTIONS OF SEEDS AND THEIR INSECT PREDATORSjPARASITOIDS 159
produce any given new acacia, r2r what is the longevity of acacias at: either site, I am reluctant (0 apply any terms that carry implications about the spatial dynamics of toe population. For all we know, three adults in the diffuse site may contribute all the acacia adults to be recruited to the population during the next hundred years.
T,.b.e" two specie~ of bruchids occur as seed predators on no other spectes of plants in this habitat. The adults can, however, be taken with a sweep net from.the foliage of the habitat at any time of year and both have been caught drinking nectar in the flowers of Parkinsonia aculeata. There are at least 80 other species of bruchids that can be reared from seeds growing within a 1 km radius of the site; none or"tnese' feed on'iicacia farnesiana seeds.
A. farnesiana bears scattered flowers from July through March, but th~,major flowering period is in the dry season at most sites.' The flowers are obligatorily outcrossed (if ~y experimental crosses made in the lowla~ds of, Veracruz in 1964 are representative) and the pods mature by the'middle of the following dry season. At the time of pod collection in the present study (14 March 1971), about 50% of the trees had dropped all their pods and on the remainder, most pods were mature. The indehiscent and heavy pods fall straight: to the ground, where they are occasionally eaten entirely by deer, peccaries and cattle, chewed up by rodents, or carried off by ro"d-ents ... - Apparently these ,animals are interested in the slightly sw~et pod wall, which may also contain a substantial amount of nutrients if we can reason from the analyses of pods of other mammaldispersed acacias (Gwynne 1969). Some of the pods fall down the cracks in the drying mud surface and others are probably washed 'a~ay from the parent plant by the torrential downpours that hit this habitat during the early rainy season: It is do~~tful if any of tbeseeds consumed by mammals are killed by digest:ive action, ~ut'rodents may chew through their hard seed coats.
The bruchids are ?~esent in the habitat a~d wil: ~~inosi! on a full-sized pod at any time of year, as shown by fresh -=;:2"S 0:1 ?ods in miscellaneous collections of full-sized gre-=n pods ~3d-= i:1 and near tte swamp in June, July, August, SeptemLe!", De:::e;'.:-,E:::--', ?e::ruary, and March. This makes the plant's behavior, of holdi:1~ the ~Jd size Qui te small until only a fevi months before i"': will IT.3"t',,;!"-=, '1"':'7 understandable. If full-sized pods Here presen"t froIT. :"":l.'Yde!" ir,g to fruit drop, the potential for total seed destru:::tion ::J' ::-.u':" -ci:_le bruc~id generations within a single acacia crown would be very high. The bruchid larvae bore through the pod Hall, into the seeds, and mature one to a seed. ?resumably there is intense intra- and interspecific competition among the two species of bruchid, as there can easily be ten times as many eggs laid on a pod as there are seeds inside. Presumably, as in other bruchids (e.g., Scheelea pal~ nut bruchids, Janzen 1971b), the larger larva simply eats -:he sr.-.aller
160 DANIEL H. JANSEN
in~dd2rs. About a month later, the adult M:~sestes'emerges throu~~ the side of t'ne seed, cuts a hole through trle pod wall (or uses some other bruchid's exit hole), and emerges. Apparently; immediately after ma~ing, the female can oviposit .within the pod crop. This suggests strong selection against asynchronous pod crops within an inci vidual's crown, and sho\.\J.d·, s~lect strongly for intra-p:)pulation synchrony if the acacia.So· are cioSe...enough to mutually infect' each other with bruchids •... .
-"~
The bruchids will also ovip~sit on'pods that have fallen to the ground, which brings out an interesting aspect of their coevolution with the acacia. When the pods are on the acacia, the beetles lay their eggs on all sides of the pod; presumably when a small object liKe an acacia pod Is fully exposed-~6 the breeze, no side is a t, substantially better microclimate than another. However, once the pod is lying on the fully insolated ground, the situation changes dram3.tica.lly.. .There ,is almost no air movement right at the ground surface, and the soil surface attains temperatures of 50 to 70°C .
. ' The female, who oviposits at night. ousk and dawn, responds by laying her eggs almost entirely on the underside of the pod, ~hich is presumably cooler' and may be slightly more humid as it is directly against the ground. For example, 25 pods on the ground under an A. farnesiana tree in the dense clump had the following number of eggs on the surface of the pods facing upward: 2, 4, 3, 4, 2, 3, 1, 0, 6, 3, 2., '6, 4, 2, 5,6, 2, 5, 6, 4, 1, 4, 0, 1, and 0 Of = 3.04); 0:1 the surface, of. th€ ·pods facing down\:lard, the same pods had 27, 12, 3S: 33,13, .29, 23, 12, 15, 5, lS, 17, 38, 15, 18, 17, 8, 13, 32, 10, 34~ 2Z~ 37, 4, and 9 eggs (i = 20.0). The eggs on the upper surface were probably laid on the pods before they fell. These numbers of eggs per pod are by no means exceptional for pads on the ground. The numb~r of eggs per pod on the tree averages a~out 8, with most values falling bet ..... een 1 and 15. However, though virtually every egg la.id on a pod on the tree produces a larva that bores dO',n-, through the ma~uring pod wall and into the relatively ssf~ seed, :~s
larv3e from eggs on -:he fallen ffiature pods have a c:'ffi =._~ -:: time penetrating the hare pod wall and the h~rc seed coa-::. ::s~e=:io~=
of drv hard pods that have been oviposited on in vitr~ s~~~ed ~~~~ once the pod has been off the tree end dried for a ::,~r"~:-" ::-.e lay·;o.,,: do not make i~ froffi ~he egg to ~he i~side of the seed. :~e~ Ce~,
however, rr,ature from eggs laid directly on mature see::s ir. ::he labDratory. :::~ is nD1: known if ei"C!',er species of Ni",:,,~e:-;~e.? is more responsible for the eggs on the fallen pods or or. -:r.e p'Jds seill on the tree.
If the bruchids were to lay these eggs on the pods ~~:'le on the tree, at least half of them that ..... ere laid near ~r,e :::;-.e of ;:~::
fall would run the chance of being killed, since at least half of them would on the average end up on 'the upperside of the fallen pc.-:. Such a situation should select for a bimodality of ovi;:'Jsition behavior. The female should oviposit inciscriminantly Gf, trJe po:
INTERACTIONS OF SEEDS AND THEIR INSECT PREDATORS/PARASITOIDS 161
sur:2:ce until about tL" time tr:,a.t the pod is ready to fall, and t~.erl s't'JP until the pod is cn the ground. It is possible that each s?ecies of Mimosestes occupies one. of these peaks.
i;i th this baqkground in hand we are ready to examine the outcome of as~ing "do the bruchids kill more A. faronesiana seeds in the defi5e-j,~p than in the diffuse clump"? The first attempt at answering th~s question. and the only one to be discussed here, was to simply collect up to 200 pods from the ground below each of 36 A. faronesiana in the dense clump and 11 in the diffuse clump. The idea was to mimic a dispersal agent removing a set of pods, with the intent of thereby being able to state how many seeds survive bruchid attack (this assumes that these bruchids do not locate dispersed pod$_Qr.seeds). The method, and t?terpretation of results, suffer from not knowing how closely the sampling time matches the time at which dispersal agents would normally have removed seeds or pods. The pods were placed in plastic bags and the beetles allowed to emerge. They were shipped to the University of Chicago and a sample of 30 to 100 pods from each bag was x-rayed within six 'leeks of collection in order to determine how many seeds were intact, aborted, contained bruchid immatures or had exit holes from them (mammography x-ray film is ideal for this purpose). There were no signs of small bruchid larvae in the x-rayed seeds, indicating that in spite of re-oviposition on pods in the bags, none of the larvae had managed to re-infest the seeds.
The results' of this sample are presented in Tables 1 and 2. The raw data are presented in full since such data are non-exiscent in the literature and I suspect may be of use to later workers as our understanding of the interaction between bruchids and seecs increases. There are a number of glarinf, conclusions to be dra",:;-" but their interpretation is far from simple. In short, in the dense stand of A. faronesiana, the~e were significantly greater n-.~:,_bers of pods per tree and seeds per po~; tte dense stand also ta~ an apl"Jarent lower ~Jercent seed mortali~'/ :/~/ ~rucrl.ids ar:'j a:"J 2~.~_:::.:,:~.:-.:
these are added, we find that the per=en: ~ead seeds i= ~~ent~~~_ in both sai:l?les. IrJ ~he :f~r~al su~mati(JL~ -:hen, -t:.rl~ /'1 • .. -f'::"~·/~f:.;':'''·>:::·'~~: in the dense clump produced neaY'ly l5 ti:::es as l~,"';-Jy via:>le see:::::: per tree as did those in the diffuse St<'JI!::i, but it Has Gsne to; :_y:ducing 7 times as ffiany pods peY' tree as i~ the diffuse s:~~~ y~:~e~ than through differential seed ffiorTality iL the tHO st~n~s. ~:~e~ey
as a member of my audience once said TG r.c:: long agG, tr.s ea::::l t~.i:.;::
is relecting the null hypothesis. The ilard thing is fif'Jring c..:t what alternative to accept. I think tha~ it will be profitable to take each of these parameters and examine it in turn in the eCG!G~7 of the acacia and the bruchid.
162 DANIEL H. JANSEN
Table 1- Statistics on pod crops of Acacia fa.Pnesiarza. in the dense stand (n = 36). For comparison with pod crops in the diffuse stand see Table 2 (see text for details).
Number of Number of % seedS:: viable seeds pods on X of ~kilfed~~ ~ seeds produced tree (est.) seeds/£od .' bruchids- "aborted" by tree (est.)
tStatistics based on arc sin angular transformation.
INTERACTIONS Of SEEDS AND THEIR INSECT PREDATORS/PARASITOIDS 163
Table 2. Statistics on pod crops of Aoaoia farnesiana in the diffuse stand (n = 11). For comparison with pod crops in the dens~ stand see Table 1 (see text for details). ,t values for differences between dense and diffuse stands are provided. all of which are significant.
tStatistics based on arc sin angular transformation.
Numbers of Pods per Tree
There are two distinctly different potential causes for che greater numbers of pods on the clumped trees. First, the clurrl?ed trees could be growing on a site in which A. farnesiana adults can harvest more resources than in the diffuse area. While t~e crowns of the A. farnesiana in the diffuse area were noc notiseably more crowded by themselves or other species than in the clump",d area, one can say nothing of root competition, soil nutrients, sail drainage regimes, drought impacts, etc. One could even argue that the rarity of adults in the diffuse area is a reflection of t~is, operating through the difficulty of seed becoming establishe~ in that micro-site. However, such a conclusion is confounded b:: :-:-.e fact that the adults in the clumped site are producing nearly 15 times as many viable seeds as are the adults in the diffuse site (though it is clear that this need not necessarily lead to a higher density of adults). Such a "resource hypothesis" would be very hard to test unless it happens to be that the nutrients in shart supply could be replaced by fertilizers.
164 DANIEL H, JANSEN
Second, the widely spaced trees could =~ directly pollinatorlir-ited, I am reluctant to accept this as c: general reason ::c,y' small seed crops, as it appears that a tree species would soor, adiust its degree of selfing to pollinator ability. However, it can ~bvi0usly be a cause of small seed crops for small fractions of the population, especially if t~ey ~arry the genetic material possessed by the majority of the populati.?%.:. a majority that is not normally pollinator-limited, ,In' tt.e case,. dlscussed here, it is quite possible that the dense clltmp, .,-hen in flower, produced a very large pollen resource, one in \;ihich it was more profitable to forage than the scattered A. farnesiar~ a short distance away.
The third possibility is that dispersal agents are removing the pods much more r4pidly from beneat~ the widely spaced A. f farnesiar~_ This possibility was appreciated at the time, and I searched the acacia crowns to see if the numbers of infructescence stalks in the crown was in approximate agreement with the numbers of pods on the'ground. It was, so this possibility must be rejected.
With the data at hand, there is'no way of choosing between one of the first two, ,hypotheses, This is so even if the inflorescence scars can still be counted. If the scattered trees are genetically programmed for life in the habitat containing the high concentration of A. farnesiana: they may well bear flowers in numbers appropriate to the pollinators that normally frequent such dense stands, and ~ay produce f1'owers in numbers appropriate to the size of seed crop that can be' produced by'the energy reserves' of a plant growing in such a site. However, "just as mentioned above for the resource hypotr.es:s, the pollinato~ hypothesis can be easily tested. One simply has ta hand-pcllinate the flowers of the scattered plants and then exa~:~e the size of"th~~iable seed crop.
There is, however, an even more confusing alternative availa~le. If the widely scattered plants are "program;;;ed for a scattere::l :>"s", v..·i th their flo;.;·e·p cr::-- s a::justed "to :::"~\-:rl "the i~U:i.jers c;~ pc:liL~ -.:s~::. ~iiat nor'~ally cc~e t: ~~e pla~:s~ 3~~ ~o T~e ~~0~~~S s~ rESE[~~~
normally ~athered by such plants, tts~ hand-?ollinat:c~ CCG:d ?~~~
a quite false :n!pressicT; ()~ wrjcl: is ~c':'ng OT... :~~ ;.2.C.:-l-:S ni~:-.- '<.":-::_~
set mere seed a~d theD C0me to vEgEt2~~Ve d~S~S!Er ~~r:~~ tte =.:lowin? dry season, or in the next se~~~e cro~~ co~~~~~~ior to ~~.~~~.
they were subjected. I w8uld have c~nclude~ ~hat t~sy wel~e p~::~~ator-limiTed, when i~ fact they ~re ~Lergy-li~i~ed ~~~ have ~~~ evolved a mechanism to "j.;:now" "'~;-.I.~;) tos) r:-lany ?ods r.l3.~..r~ been Sf:"": ~"':. ~
depend on the low pollinator accivity to de"LerF.,ine tr.is, ho'"e',sr, I personally doubt that wild plants become quite so dependent o~ external factors with such higt, a pot en"': ial variance as the nUj;.~sr
of bees to arrive at a flowering tree from year to year. I s[,o'.;.l-: at this time point out ttat no" one has ever recorded this para~e:Er for a tropical plant (and I have never seen such infor~ation for ~
mid-latitude plant), Ey the same token, it seems highly unliks:1
INTERACTIONS OF SEEDS A~:D THEIR INSECT PREDATORSjPARASITOIDS 165
that a tree would reI:.' on hav~ng the same amount of resources dccumulate each year 'ith which to make fruits. Perennial ~lants must therefore have some interpal mechanism to decide how many pods can be set with the reserves of a given year, and they must do it early in the progression of flower primordia to bud to flower to immature p.od,
Numbers of Seeds per Pod
This is' a most perplexing parameter. Numerous selective forces potentially affect the phenotypic trai"':: "numbers of bits into which the total seed crop is divided".
" -.- • ... . f As the number of seeds per pod decllnes, the amount of resources
(pod walls, pod sugars, pod pulp) expended per seed increases. I hasten to add however, that the amount "spent" per seed dispersed need not necessarily increase. This depends entirely on how the dispersal agent community as a whole responds to small packages, large packages, or some size discribution of packages. As the actual animals (how many of what species) to arrive at a given tree depend on that tree's exact location, the tree genotype can be molded only toward some sort of "average" pod crop ane dispersal. The ",rariance in seeds per pod that we see among the A. farnesiana in each of the sites may be the outcome of direct selection for variance, or the outcome of various selective factors tugging in different directions, or some more horrendous combination of the two. The diffe~ences between the two sites can thus be a result of active selec~ion, if each of the two habitats represents a distinct deme. I think this is not likely, though genetic differences between such prox~-;;-.ate
portions of a plant species are certainly well known. On the other hand, the difference in mean numbers of seeds per pod coule a:so be simply the result of the two factors discussed in the yr~·,rious
section, to wit, either the trees are starving or not eno~2t ?ollen grains are arriving on the stigmas.
Unfortunately, we have no idea if tropical trees var~ :~e numbers of seeds per pod as the resources available ~o ~he tree vary. However, it is quite conceivajle that the ability o~ the :ree to attract dispersal agents changes as the pod crop size dec:~~e~, and with a different array of dispersal a£,;ents arriving, the :,~,-::>.al
number of seeds per pod should be di~ferent. It s~ould ~e ~::ed that the number of seeds per pod declines in two ~ays wi:~ ;. farnesiana. On the one 'hand, there may be fe;.:er seec:s iT, ~ ;')d of the same apparent size, and on the other hand, there ~ay ~s s~aller
. (shorter) pods. Both cases occurred in the samples at ha~~, ~ut such a distinction was not made in the original census. I~ was quite obvious, however, that certain trees produced almost entir :"'y very short pods, and other trees produced almost entirely very ')~g pods; this difference in pod length is largely responsible for L e
166 DANiel H. JANSEN
d~f~e~ences in mean numbers of seeds per POQ from tree to cree, "wei tllis in turn means that the Hidely spac,c'J A. far>nesiana trees ~ore often had short pods than did the A. farnesiana ~n the dense c~ump. This conclusion, incidentally, implies that it is reserves that are in short supply rather than pollinators, because if it is only pollinators, there would be.:no reason for those flowers that L.,m'e T=,llinated to produce smaHi~ds. Turthermore, if it were the nUr.1ber of pollen gralns arri"ing at-the stigma that was setting the numbers of seE;,ii:ls' at a lower level, then we would expect a high variance in the si~es of pods within a crop, rather than for some trees to have many short pods and other trees to have many long pods.
The bruchids have to be considered as part of the selection for numbers of seeds:_per pod as well. _ 'Lhey figure in two ways. f First, if fruit size is important in the rate at which seeds are removed from the tree, then a factor in selection for an optimal fruit size will be that the longer the pods stay at the tree, the more seeds will' be killed by the bruchids. Second, it is quite
"possible that there is an optimal distribution of pod sizes (or seeds among the pods) that minimizes 'the number of seeds actually gotten by the bruchids before dispersal.
Forgetting the dispersal agents and developmental impossibilities for a moment~ we may imagine at one extreme the seeds contained in a small number of I to 2 meter long pods. At the other extreme, each infructescence could be festooned with 2 cm long pods, each contai'ning oniy a' c'o~ple of seeds. If'the ends of the pods were done up econ6mically, the many-pod morph might cost little more than the long-pod morpho There is no information that would allow me to determine which of these two extremes might lose the most seeds to l1imosestes s,a~laei and M. immunis. These two beetles lay their
_ single eggs at short intervals along the pods and there is no obvious way that one pod structure would result in fewer bruchidkilled seeds' than the other. If the paj were long, t~en t~e ~rilchid could simply w~I~'along it laying eg~s. When the pods are t~~ided into many small pieces, the bruchid co..:_:' walK dov:n o;-,c;, :-,s;-, -:.c- :r,s next, and so on.
It is possible tt.5t ther~ are ex~s~~~: ~2ctors c;s~~-_~~. A bruchid on the two di~ferent pod types cOill~ be exposs:' ~~~~E differentially to preja.-c':'Y's. :;: cann-ot, '-',o'.-:e';sY' , sec; :-!~~-.-.' r,:.S ~::
these tHO contrastin[: pc-:::' types coule :=C'.Tor ;-;j·;;-,enoptc;,;,;-, ;~""2S~'_C:::'s
(to say nothing of the :=5Ct that : ca~ns: reco~nize -~~~ :~ ~~~:= ~e
selected for even i~ it did, provided -chat thebruchid ~il:eG the seed before the parasitoid killed the ~ruchid).
Finally, we must add that the physiological and dEvEl0p~ental costs of the two extreme types of pods s~ould be considered. There are numero~s legume pods with many more seeds per pod thaD th5t of A. farncsi-ana, and likewise, many with far fewer. This does not
INTERACTIONS Of SEEDS AND THEI,R INSECT PREDATORS/PARASITOIDS 167
fr.ce us from the problE:n that as. the selection pressure for a larger nu;;,;-;er of seeds per pOd slides upward, the physiological cost: of responding to it also increases,' through such things as having to increase the average number of pollen grains (or the size of the polyad) that hit the stigma, which may in turn require changes in the flo~al ~ttra~tion to attract a different set of pollinators, etc.
Percent Seeds Dying
This apparently simple parameter has turned out to be the most ccnfusing of the lot. The per cent abor~ed seeds is significantly :~ower in thewidel~ scattered acacias but: th~percent aborted seeds is significantly lower in the widely scattered acacias but the percent unambiguously bruchid-killed seeds is significantly greater to a degree such as to exactly complenent the abortion percentage; in the dense acacia site, 21.30% of the seeds were viable and in the diffuse acacia site, 21.35% of the seeds were viable.
The most attractive hypothesis is that the plant aborts a seed when a bruchid larva enters it, if seed development has not p~ogressed past a certain point. If this hypothesis represents the real world, then it suggests that either the pods in the diffuse stand mature earlier or that the bruchids find the pods later in ,their maturation cycle. Whichever the case, the bruchids kill the same number of seeds in the end in both stands. This simply suggests that from the bruchid's standpoint, the acacias in the diffuse stand are not far enough apart to cause a reduction in seed predation, at the smaller sizes of seed crops found in the diffuse stand. If the average seed crop per acacia in the diffuse stand had been the same as that in the dense clump, then it seems reasonable tha't there might have been a lower percent mortality in the diffuse stanj. This assurr.es that a single tree with a large pod crop is not as conspicuous (in its odor) as is a clu::lp of A .• .t'cY'iles:'cna. i.n ~::-'..:i t.
A second reasonable hypothesis is ~~2t tte bruchids ~ay ~~~rc~. (}Ul: pods or trees wi~h a 10\'; nu;-:-;~)er of o:J0r::eG seec.s. ~-h':.s ~.::_:",c2r's
less likely, but should not be discounted. •. t:~is should :_r~ out to be tbe case, then it sL:ggests that the VE:r~l loT're number s~ pods present in the dense clu:r.p hati'tat have sati2'tec: 'the :::,ruc:hi:.:.:, 1
ovip8sitional ability to SODe dsgree. ~L ~ls0 re~uires aG ~:~~:2G-
the widely spaced plants. ' A lower physiolo ical ator'ticn r~t": '::'n the widely spaced acacias would favor the i ea that they are poll'::'nator-limited rather than resource-lim ted.
168 DANIEL H. JANSEN
CARRYING CAPACITY
It :s tempting to intuitively consider the number of bruchids generated by the host population's seed crop each season as a prim~~y parameter in determining how mqny bruchids arrive at the seEd crop in the next season •. This, however, assumes that the adult nortality between seed _srops·ta~~some fairly consistent form each veal'. and may be practlcally consioered to be independent of host ~nd bruchid density. The latter assumption requires that anyone species of adult ~ruchid makes up such .a small portion of any general predator's diet that even if the b"eetles were to be abnormally co~mon, it is unlikely that a generalist would temporarily specialize on them. Furthermore, there do not appear to be any specialist (arthropod or vertebra~e) predators op.adult bruchids moving in 4he habitat at large. However, there a~e two things wrong with such 'a simplistic view. First, the adult bruchids (at least some species) do take nectar and pollen from hosts and this may be a resource for which they' compete as adults (nectar and pollen fed to laboratory bruchids increases both longevity and fecundity). Second, the adult beetles may congregate in particularly desirable sites in the forest and these high density points may attract predators.
Resource Size of Adult Host Plants
Mimosa pigra (Hirnosaceae) is a common low shrub in marshy cattle pastures of Guanacaste "(and much of lowland Mexico and Ce~tral Ar:;erica), Hhere its'seeds are attacked by the larvae of Acanthoscelides pigY'ae, a'smail bruchid host-specific to M. pigY'a. The natlJral habitat of this plant is marshy areas along frequently flooding ri'Ters and ,op ,piyer banks. At Palo Verde, mature pods ma:{ be found on the plant from late July through r':arch, and the first flNters appear at the beginning of the rainy season (April-:~a'l). A neH crop bf flowers is produced each day. The adult bY"jc:h :ds ,,-.oJ' be ~ound on the" flo;-lers at davm, ap;)earing to feed on tbe ::;sl::'e".
To exa~ine how ~any ad~lt beetles are associated ~i~~ ~ ~~~:.~~
ing plant, an isola:ed ".:. pigY'a b1.:.5;-, Has selected ae: e:;-,e S::~-:', -=~,:
of ThE: pasture im;ne~icTel::l to the sO'~-:=h o~ the Pale Ver':>-;: :~€~~
Sta-:ion ment:oned previously. The Learest other U. pigY'~ ~_s~es Here 212 meters away, and these bushes for~ed a dense patc:~ o~ aboue: one half hectare in area (i.e., a very large poteLt~-=: so~~:e
of A. pigY'ae). On the mornings of July 10 to 12, in 1971, ::: collected all the beetles on all "the rloHers, 'for totals 0~ ::':'L,
289, and 314 adults. On the mornings of July 13 to 16, I caush~
only 26, 19, 24, and 25 beetles. On the mornings of Julv 17 ~o 22, I caught the beetles but released them afterward at t;,e ce~Ler of the bush. The numbers were 24, 41, 72, 69, 136, and ISS. -:~
rained on the nights of July 19-20 and 21-22, the two nig~ts
INTERACTIONS OF SEEDS AND THEIR INSECT PREDATORS/PARASITOIDS 169
preceding mornings Ln which- the teet Ie numbers did not increase. I interpret these results t9 mean that the shrub had a carrying capacity of about 300 beetles, and the "island" had an immigration rate of about 20 to 30 adult beetles per night, except when it rained _ (This' plant would produce a minimum of 60,000 seeds during this fruiting season.)
-- ':.~,,-
It~ seems 'evident that in this example, conditions were crowded e~ough somewhere, a good two months after the last seed crop, for the beetles to'be migrating among the plants that were to be their oviposition hosts as well as the adL.:..t food hosts. In migrating, they expose themselves to a wide variety of mortality factors which can in this sense be viewed as density-d~pendent. Furthermore, once they' arrive at a bush, it appears that the carrying capacity of the inflorescences is very substantially less than that of the bush as a whole, again indicating that conditions may be crowded on the inflorescences.
It should be noted here, however, that M. pigra is exceptional among the bruchid larval hosts in the deciduous fores't in that the adults are abundant on the flowers_ Adults are occasionally taken in sweep samples of general vegetation many hundreds of meters from M. pigra plants, but they have not been found on the flowers of any other species of plant.
Concentrations of Adults
As reported in a recent sweep sample study of seasonal changes in insect density (Janzen 1975a), large numbers of small beetles may become concentrated on the shaded understory foliage beneath individual evergreen trees in the deciduous forest during 'the las~ half of the dry season. A substantial number of these ~eetles are bruchids anc Heevils that pro;,:,ablv e:~ierse::: :'Y'J::. s",,,,:'::::; eClrlier ir, th~ cr~: season. It is conce':~,.rc:·.l-::. ~~at a:--, :::z-=-€::-·-::J:-.~~~·.~ =-~:-'~E' . .
num~er of tru:hids c~ one s~E2ies :o~lj cc~~~~~·~~e ~~ ca~s~n~ sIT!all insec-t:iv:)rous :'i.rds t') CC\~;C€:-}--::!-'o.-:'~ ~_!l'=:':;· -:= ____ '~2~-.::-~;- ac-:=i",,"i:\' at trlese S~~-2s. Hav.re?er .. if this 1::: tLe ':-~~~~ ~--:: ::: C.:._s:: :.'~;SSi2~-::
for 2Tl ou-::~,rec.:,<: of 2:-<,· other s;;'2ci<=s ''J"f ':r:=~c.:: l:-:J -=--:' T!-l~ S'e:::-.'2, ,,;:~.:.-:.:.
may be one of The ~ulTitude of way~; in which :~le ~?r)er li~its o~
insec~ species rich~ess in such a ~1~jitaT 6Y'~ SET.
A core aspect of the theory aevanced earlier (Janzen 1970) is that scattered or dis?ersed seeds are less likely to be found by seed predators than are seeds in the cluster re?resented bv an undispersed seed crop (either still hangin~ in the tree or merely fallen to the ground below). That such dis?ersal is effec~ive has
170 DANIEL H. JANSEN
been shown with seedlings of a deciduou~ forest legume (DiocZea megacarpa, Janzen 1971c), nu-+:s of a larrr~ palm (ScheeZea rostpata, wilson and Janzen 1972) and seeds of a sterculiac~ous tree (StercuZi~ apetaZa, Janzen 1972). However, it should be recognized that, just as with the dense and diffuse elumps of Acacia farnesiana, there are lower lilTllts to the' distances that seeds can be separated and show a reductio!!. in see·d.~~dation. An example is offered by Spondias mombin (Anacardiaceae). .
J .-
S. mombin~is a CODmon tree in the deciduous forest around the Palo Verde Field Station. It p'roduces large numbers of ripe ~ruits fr'Om the middle to the end o~ the dry season, fruits that are eaten entire by monkey:,> (and probably by birds and other nammals). The large multi-seeded . 'Woody nuts are -then defecated onto the foY'st floor somewhat cleaned of their outer pulp. At this time, a .arge undescribed species of AmbZycerus bruchid searches chern out and glues .• a single egg on each one found. The larva bores in and eats all the seeds. However, with the present density of dispersal agents in this forest, the majority of the fruits end up rotting on the ground below the parent tree. Once the fruit has rotted off, the Amblycerus oviposits on them just as if the fruit had been peeled off by a dispersal agent. This concentration of nuts beneath the parent may be vie·,.;red as either a "bruchid sink" that lowers the number of beetles that find the dispersed nuts, or as a "bruchid generator" that probably provides 90% or better of each new generation of ~rucb~ds. Incidentally, this is another way that the presence of seed dispersal agents can clearly affect the density of the se~d predator, which may in turn influence the density of adult trees.
To .examine the intensity of seed predation directly below the crown of the S. mombin trees and compare it with that at the outer edge of the se~si shado'" produced only by fallen nuts, about 100, 1-2 month Q~q.nuts were collected from each position from beneath 12 larse S. mo."noin Fr'Y.·:i:J~ in the fGrest about 4 j.-;;, north."les--c o~ the ?a:o Verde ?iel~ Sta~ion. The densiTY c~ n~~~ ~an be ~s ti~~ 2E 20[' ~n a s~~are meter, di~ectly te~eaTh the C~'~~:~, jut :h~ aVera[£2 is a!:!':J"liL 2:; r-el-' sC;'Ja;:-'~ me"tE:r. The :::erce:-.: :.. ~ nUTS ...... '.:. --::.r. A~bluc~r~lS exi~ heles 2T t~e base ~~ the trse an~ ~~~ectl)T ~:l~er t~:e ou:er e~fe of T~le cr8W~ (4 to S rJeters ~rOTh l~.S ~ase o~ ~~e Tr·~e)
40 (~ = 28). There is DO hin~ o~ a difference b2:~22n the ~~o ~e~~s. r~rther~ore, IT the co~parison is made pair-wise, ~~e number of exits from nuts at the base of the tree agrees closely with the n1.:mber of exits from nuts under the outer nargin c~ -::he crov..-n. h. careful dissection of 523 nuts ~rom below three S. ~~Mbin show that 25% of the nuts lacking exit holes but with Lr~chid eg~s on the~ h~ve their seeds killed by a bruchid, but the ~ruchid ~~ed before it could emerge. The remaining 74% of the nuts with eggs
INTERACTIONS OF SEEDS AND THE!R INSECT PREDATORS/PARASITOIDS 171
on them had aborted or r·:--ted seeds inside (cause unknown, Dut cou:'.d "'ell be that the bruchid larva attacks the seed and then dies while
. too small to be seen later on in tne decomposing material, as happens with bruchids in Sc:heeZea palm seeds). If we add the nuts with eggs on them to the nuts with exit holes, as an absolute measure of how many nuts, wer~ at least found by an Amblyc:erus bruchid, the respective .per£en-t·ag~are 36, 50, 55, 60, 60, 66, 50, 27, 46, 71, 36, and 45 (X = 5~) for the nuts at the base of the tree, and 41, 49, 49, 58, 19, 69, 46, 43, 30, 61, 26, and 46 (X = 45) for the nuts under the outer .edge 'of the crown. The differences in the means ~re not significant. To understand why th~ remaining 50 percent of the seeds did not generate a large seedling shadow beneath the p~rent trees, I should point out that of the 5~3 nuts mentioned above~ only 4 cOntained a viable seed (4% of the 102 seeds that had neither AmbZyc:erus exits nor eggs on the outside).
HOST SPECIFICITY
In the Guanacaste deciduous forest, the insects that breed in seeds show amazing specificity. I should add that I have seen nothing in other tropical vegetation to suggest thaL this is a phenomenon peculiar to Guanacaste deciduous forest. Detailed documentation of this specificity is still in progress, but it seems reasonable to present a preliminary view here as there is no indication that completion of the study will drastically change the picture (the data given here are a refinement on the first approximations menLioned in Janzen 1973a,b; 1974b).
In a flora with about 300 to 400 species of plants (excluding grasses) with seeds large enough for a bruchid or other see2-eatin& insect to develop in, I can state with certainty that at least 59
.species have no pre-d ispersal seed predat ion by insects, 5'j s;:,ecies have one or more species of bruchids, 6 have ',;eevils, 2 haole c.era::.::":;cids, 2nd 5 have mo-;:h ::'ar-;ae (ezclu-:'i:-.? seed cj-,dlc~ds). ~,. ~ .. ':l-::l:'~ .• at leas"t 20 other Sp~CiES c>f t)~'(..lchic.~ ;,;::ve ::'~J:::~ collt2-=-:tec. ~:-. :;-;.~.:c:.
around the fores-:: bu-;: their hosts are u:-knowL. This ~e2LS ~~~: 2
rninir;,L;.~;, of aboll~ 25 to ~,3~ of all t:rl~ ;;lant SI"Sr: i es .:i:l tt:~ '. s ~ =- ~ L -=._= . . forest s1J.f~er cc;~siceyG.:~~.~ :'=ro;-:-; F-.y'e-c':s~ersc::~ ~-::ec ~re,j6~~::.
is O~Je ~.a~or source o~ ~~~-jis~ersal :ee~ pr~~eticn by ins~~ -: t~~: is n01. considersG her'e; ::,~z,s. (;">2rrlip"':~::-'-3) 1:a:v:e c ~/o:::r:.- 2.c.~!;e:. _.~:r: =-
develo;:inp; embryos of cercain species (These are often recoy:,,;'~ cs aborted seeds) and wee-,ils (Curculionidae) develop in flowey :"uos .of many species. We can ask two quesTions of this inforDa1:i0~. How are the bruchids dis~ributed amon~ the plants, and how ay~ the plants distributed among the bruchids?
Among the 88 bruchid species whose hosts are known witr.
172 DANIEL H. JANSEN
certainty to date, 73 (83%) have only onp host species. AI! the .. eev~ls and cerambycids have only one host species. Thirteen species of bruchids (15%) have 2 hosts (e.g., Gibbobruchus guanacaste in Bauhinia pauletia and B. ungulata, Amblycerus championi in Cordia dentata and C. panamensis, Acqnthoscelides kingsolveri in two species of Indigofera, Ctenocolum tub~rculatum in Lonchocarpus eostaricensis and L. minimiflorus, .and Megac'e""trus leucospilus in Ipomoea pes-caprae and I. fistulosa). One bruchid has three hosts (Ctenocolwn crotonae in Lon~hocarpus~costaricensis, Lon~hocarpus minimiflorus, and Piscidia carthagenensis) and one' has eight and perhaps more (Stator limbatus in Pithecellobiwn sar,;an, Pithecel.lobium du'lce, two species of A'lbizia, two species of Lysi'loma, and two species of Acacia). All indications ape that location of the hosts of the twenty or~so un-reared species "wiil swell the -"one- host only" category.
There are undoubtedly a number of components to the answer to why these ·seed predators are so host-specific. Certainly it is not the mere outcome of segregation by habitat; with careful collecting, as many as 40 species of bruchids may be reared from the seeds in a five hectare plot, and 22 species were taken in one forest understory sweep sample of 800 sweeps (dry season, Palo Verde, study reported in Janzen 1975a and beloW). On the other hand, in over half of the cases where a bruchid occurs in two species, the two species of plants are quite definitely habitat-segregated. A conspicuous example ~s Megacerus 'leucospi'lus in the seeds of Ipomoea pes-caprae growing on ocean"dunes and in the seeds of Ipomoea fistu'losa growing in fres~water marshes along the Rio Tempisque.
Competition among bruchids within a species of plant may contribute to the,host specificity. If we regard those s~e:ies with bruchids as' in some sense available, the bruchids ap;:'Eoe:.r to be quite evenly distributed among them; of 69 host species, 5S% t3ve one bruchid and 35% have tW8 bruchids, while only 7% haVE: j ~ruchids and 3'1; have 4 brtte-hids. T~is conceneration of 90 es of tr;", ~_l"::HS in the 1 or 2 bruch id ranr:e ii71pli es t ha t i f -::~s nLl.IT:~ers ri 2:;".; :, '~T-=~ :: .... :J
species., int:er-bruchid s;le2ies CGT:"lpe-:':'-;::'or.l G-3.:l ~ec0::.-= "<:;:'," SE:vere. In ell C3ses o~ 3 or 4 truchids ~er to~t plant spec~s~. at le":st 90~ of the bruchids rea.red from seec sa;np~es belor,g LO t·,:':.- ::.oec':es. Suct COTIgetition ~ay be select~ng in fa~or of specia~:::::: by selecting for maximal efficiency on a given host.
The most obvious can6idate for a ~alor cause v_ s~ecializa-
tion is the need to specia:ize to overc~~e ~he che~is~~, ~e~5Vloral and morphological peculiarities of the host plant. :::;; :;~,0rt, one is forced to postulate that as the plant becomes beeter ~~~~~Ged (selected for by past generations of bruchids), the ::"~~'_le has to be progressively more specialized to stay with its host. 7he more specialized it is on one host, the more difficult ie -:;L'j"ld then be for it to also be adequately specialized to attack a =~~0nd host. Such incompatibility ca.y take two forms. (1) The cri-_~c..al traits
INTERACTIONS OF SEEDS At~D THEIR INSECT PREDATORS/PARASITOIDS 173
of the attacker of twa host species may be chemically difficult to possess simul taneousl:' . For example, the f2:.zyme system that Lrear:s do~~ the toxic secon~ary compounds in one species' seeds may be biochemi~ally incompatible with the enzyme system necessary to break down the secondary compounds in the seeds of the second host species. Again, to. locate· the pods of one species, the female bruchid might h~ve to . .s~arch in moist and dark micro-habitats (e.g., Spondias mo,'7/bin nu{S""being sought by AmbZyceY'us on the forest floor) and in sunny and "dry habitats for the othe~ (e.g., Combretum farinosum fruits being sought by AmbZycerus high in a deciduous forest canopy). (2) The critical traits of the attacker of two host species may be energetically (or nutritionally) difficult to maintain simultaneously. For example, ·it may be energetically very difficult for Caryedes br~siZiensis to have the enzyme system to 40th deal with seeds with 5 to 10% canavanine in them (which it does 'when feeding on DiocZea megacal~a, J~nzen 1971c) and seeds with a 5 to 10% concentration of L-DOPA as found in M ... cuna andJ.eana (Bell and Janzen 1971), which gro,,'s in nearby habitats and has pods and seeds similar to those of D. megacarpa.
In the previous paragraph I was careful not to imply that any combination of attacking abilities is impossible in any absolute sense, as it is clearly not. Stator Zimbatus is the most aDazing in this respect. It is clearly a specialist at laying its eggs directly on the exposed but undispersed seeds of four genera of mimosaceous legumes. Associated with this, however, it must be able to deal with the alkaloid pithecolobine in PitheceZZobium saman seeds (Magnus and Seaforth 1965), the uncommon amino acid albizziine in A Zbizia seeds, and probably other secondary col..pound s in LysiZoma and Acacia seeds. In all the other cases of 2 or 3 hosts being attacked by one bruchid, the females oviposit on the intact fruit (and the larvae bore through the fruit wall to ~he s'2eds inside), and in all cases the fruits of the two or three hosts are very similar in shape, texture, thickness, size, and ajor. T~e Gair of hosts are also congeneric in all cases bu-:: one (Lorc'3i1OC:CZY'-;/i.J.3 a:-.'_ P{scidia). Centl2r 6.1:0 Jorlnsc)T'l (197 L ) r.av~ :e::ent':">r =-=;~ss~=- -::-.'7: fruiT wall as a rr~2~or barrier to ~ruc}lid En~ry, an: :~ese !'~=_::~ sup~ort t~at e~?hasis.
7~~re are ~ive ~evs to ge: at the role o~ seec s~Je~.ls~~; ~r
this extreme host specificity.
(1) We c~n as~ how readily th~ ~ruchi~s traris~~r G~~S ~~:r~duced s?ecies of plants." I have o~ly one e~:a~ple :~ CQ~~Y'~~'l~~, thae of the shrub Cassia aZata (which may be native 'to Central America, but is certainly not wild in the Guanacaste decidu~~s forest). C. aZata in gardens very rarely have an 1:.~0ZlJcerUs anc: a Sennius in their seeds, bruchid spe~ies that normally live in one or two species of her::-aceous weedy Ca8sia that have pods ·ver:; different from those of Cassia aZata in superficial aspect. Since
174 DANIel H. JANSEN
the native and introducec Cassia probably ha'le very similar ~eed chemistry, this example is 0:' very little u'.:e. In general, r,owever, it can be stated that none of the numerous introduced legu:nes (e. g. , DeZonix regia, CaesaZpinia sp., Erythrir~ spp., Phaseolus vulgaris, Lathyrus sp.) in Guana::aste have had indigenous bruchids move onto them.
,~ (2) We can ask wh~t happens when a wild female beetle acci-
dentally oviposit§. -on the "wrong" host. The technology of this situation is almost impossible since the only time that a record will be obtained is if the bruchic'larva survives. (For the most part, bruchid eggs on the fru:t or seed cannot be identified with ease.) However, in all the tens of thousands of bruchids that have been reared to date :in .. this study, 1. baye only one unambiguous ca~e of a bruchid being reared from seeds of a wild plant other than its usual host. At Finca Taboga in southern Guanacaste, R. Carroll found a pile of howler monkey feces cO~Taining nuts of Spondias mombin (Anacardi'aceae J and the sil7lilar-sized seeds of Eugenia salamensis (Psidium Y'ensonianum; r~yrtaceae). The site smelled strongly of S. mombin fruits. There were 78 S. mombin nuts with Amblycerus eggs on them and.6 E. saZamensis seeds with one Amblycerus egg on each (out of a total of 354 E. salamensis seeds and 197 S. mombin nuts). During the following months, the Amblycerus emerged fraIT. 71 of the S. mombin nuts and a single adul! emerged from each of 5 of the E. salamensis seeds. A bruchid has never been reared from <lny other of tnousands of E. salamensis seeds, though an undescribed weev il· kills about' S06 of large seed crops.
l3) A s~~ll al7lount of data has been gathered abou~ the o~tcome of offering a fel7lale bruchid a series of seeds of wild hosts :'Y"':)::i
her !1abitato.n,yhich to oviposit, but this method has nat tee:-, exploited fully owin~ to the difficulty of maintainin~ ~ost ~~:~ bruchids as laboratory colonies. Mi.'71osestes salZaei f"';:.alc:s :-.",-':e
been fo',md tD oviposit on almost any laq;;e smoo-::h rC'l~rlS s'?'?~ ~::: -:;-'''''C; are Acacia .. fal 1 ,;:-;;;·s{C;J2a seeds pr'e~~nt: jrl -:':-le sar:-:e c~:-(t;:~~:-.(-:~ .. r __
(containing canavanine). Only the last mentione~ sp",~~es ~~ ~~~~t has a bruchid in its seeds, but all occur within the ~~:g~~ ,~~~'"
o~ M. saZZaei and nane have pod walls thicker or notic"'~~~j :s_g~er than those of Acacia farnesiana.
(4) A descriptive analysis of seeed secondary co;-;,pc:>un:i cr_~.~"'L"::
INTERACTIONS OF SEEDS AND TH~IR INSECT PREDATORSjPARASITOIDS 175
over an entire habitat "'0''';' ~ prob.ably be helpful, but not definitive. I f all the seeds in a hc::>"< :it were found to have similar or the same
.secondary COTllpounds in the~., i L WQuld be unlikely that the bruchids would be specific on account of seed chemistry. This analysis is still underway, but .it is clear that the diversity of secondary compounds is:at the level of about 5 different major constituents for each'lG·*cies of seeds. In addition to the major compounds occurring at~~o 10% concentration, there are often others that occur at lower concentrations, to say nothing of the concentration of lectins, heteropolysaccharides, endopeptidases, and saponins, all of which occur in bruchid host (and non-host) seeds, are toxic to certain bruchids, and have not yet been extensively sought in seeds of Guanacaste legumes.
(5) Secondary compounds from seeds can be incorporated in bruchid artificial diets, and such a study is currently underway in collaboration with E. A. Bell (nitrogenous secondary compounds) and I. E. Liener (lectins or phytohaemagglutinins). I chose Callosobruchus maculatus (stock obtained from the USDA Stored Products Insect Research and Development Laboratory, Dr. E. Jay) since this bruchid is easily reared in the laboratory and feeds readily on black-eye pea seeds (Vigna unguiculata) which have no alkaloids or uncommon amino acids. This bruchid may be viewed as biochemically/physiologically naive, and therefore a test of putative seed toxins on its larvae may represent what happens when a wild bruchid ~ays its eggs on the seeds of a non-host.
The logistics of testing are very straightforward. A Wiley mill in the Uni versi ty of llichigan CoEege of Pharmacy is used to reduce dry black-eyed peas to fine flour. For control seeds, this flour is used to hand make cylindrical tablets 13 mm in diameter and 7 mm deep with a mechanical pill press (also from the College of Pharmacy). The females oviposit readily on these pills ~hen placed in jars containing stock C. maculatus cultures (they also oviposit on just about any other smooth surface). All egrs but ~en are removed from the pill, S~ 2S to ad4~s~ ~he nu~~er o~ develoring larvae to where ~here w~l~ be very lit~le intra-s?ec~~~c cs.:1,;:.E::ition. P.n averave o~ five ~2 si;: 2c~lilts E::--~.:::::-·~~e :=1"'C!;" "t~le :.~:-.
, '-1 Ffr~ on stieG a pl~ . l"I1'2 ~i.:Ll is made, the secc:-J:::'2.r~l Cv~,;18\_:.n,j is r;JiXf2~ in a cry PQw~.::::r
form with the flour. To da~e, ex?eriments have ~een with 0.1, 1, and 5~ conc~~Tration of ~~condary co~.pounds~ as ~his is :tA~ ra~~~ of concentrations at v:i1ich secon:lc.r:, compounds nor;:-,ally C::::Clr .if. seeds in nature. The pills plus eg~s are culture~ at 20~ ~~ a~ roo~ temperature; such care is not necessary for intact seeds but the pills dry out much more rapidly than do intact seeds.
The results to date are inco~plete in coverage of the secondary compo'Jnds found in Guanacaste bruchid hosts and non-hosts, but provide some interesting previews. For example, lectins or
176 DANIEL H. JANSEN
phytohaemagglutlnlns, compounds well known ~0 be toxic to mammals v;i,',';-J taKen crally (3S well as intravenously' (Liener 1974) have been shown to be toxic to an insect seed precator for the first time. r~:,en black bean lectin (Phaseolus vulgaris from Guatemala) is incorporated in the pills at 5% concentration it is lethal, and only two beetles emerged from five pills·.with 1% concentration (at 0.1 9",
er.Jergence is indist ingu}shable-.~ the controls).
Alkaloids ineOrpcrated in the pills may be even more lethal. Colchicine kills ~~l larvae at 0.1% concentration, and caffeine at the same concentration produced onty four adult beetles out of five pills. Caffeine is lethal at 1% concentration. It is appropriate to add here that the beetles that are produced on these marginally texic diets have not-been checked ·for_reproductive abilities, but~ may well have reduced reproductive. fitness.
Uncommon amino acids have been most extensively tested to date, and have produced the most confusing (and interesting) results. The following uncommon amino acids from seeds are lethal at 1% concentrat ions: L-DOPA, B-aminopropioni trile furr,arate, L-dj enkolic acid, N-methyl tyrosine, B-cyano-L-alanine, and L-IT,imosine. All of these also show very severe reduction in emergence at 0.1% concentrations. A ~oderate reduc~ion in bruchid emergence was obtained with 1% concentrations of canavanine, albizziine, D,L-2, 4-diaminobutyric acid, y-rr:ethylglutamic acid, and m-carboxyphenylalanine. These and others are being 'tes,t.ed ,ar.the 5% concentration, and car.avanine, y-methylglutamic acid, and m-carboxyphenylalanine are lethal at these concentrations .. However, it should be added that D,L-pipecolic acid, L-homoarginine, and S-carboxyethylcysteine at 1% concentrations had no visible effect on the bruchids. In short, the effects are highly variable, Concl';rrtl."ation dependent, but largely toxic at some level representative of that found in seeds .
• • f... secorid f9!'.rri'of ~ontro~: e~pecially a:;.,?rC!?::-~ate. to the que'sticlD of wnether unco~mon amIno aCIds In seeds are ~ere~y nltro~e~ stor~~e c~mp8~nds, or defense mec}jan:s~s, (·r bo~h (S~~ ?~s~~thal 1~72 fcy' an eX~;'i;::'le of ~:>Th i:-l Ca.rt{lva2iQ seeds wit:r~ c;:~;~c;-.r~:-.:rje :'n tL~ill), j::
latus. At the 5% co~cen:ra~ic~, we have fc~~~ :~;e ~-iso~ers of Tyrosine, try~Tophan hydr'oxy~roline, aspar~~c 2~~~, cystine, and methion~ne to be lethal, the L-isomers of c7s~e~~e, leucine, isoleucine a~d histidine to be ~ildly depress2~~, ~~~ ~te L-is~mers of arginine, elanine, glut~nine, tllreonine, ~~'~:_~~~, [luta~.~c aci~~
p~enylalanine, asparagine, and valine to hav~ DC! 0bvious ill effec~. It "as expected that a SQ6 concentrat ion woulc: ;-,ave no effect on C. macuLatus, and so now the toxic protein a;;.i:-10 acids are being tried at 1% concentration. These results er:.;:>hdsize t:-lat simple n"-ltrient imbalances amonp, species of seeds c:,,\,;.lC ::'e enough for toxici ty. It is of part iculdr interest her", t~,~t ty·yptophan, cystine and methionine occur in exceptional:y low concentrations
INTERACTIONS OF SEEDS AND THEIR INSECT PR,EDATORS,'PARASITOIDS
," Llack-eye:::' pea: (e.g., Johnson and Edyrrlond 1364, EVens and bandemer 1967, Sevilla-Eusebio et al. 1968).
HYPERPARASITOIDS
177
·-Ady.4iscussion of the coevolution of parasitoids and their ;-losts, 'Or predators and their prey must take into account their hyperparasitoids or parasites. If He view the bruchids in the seeds as parasitoids, then the parasitoids that they have may be vieHed as hyperparasitoids. In the Guanacaste dry forest, all bruchid hyperparasitoids reared to date have been Hymenoptera. However, just as one does not expect all arthropod parasitoids to have h)~erparasitQi~s, not all bruchids have byperparasitoids. In fact, none have emerged from the seed collections for at least 35% of the 73 bruchid species mentioned earlier. From another 38%, only one or two individual hymenopterans have been reared from collections that produced hundreds to thousands of bruchids. Even in the remainder, there is little suggestion that the hyperparasitoids are taking more than 10 to 20% of the bruchids (I hasten to add, of course, that such a low percent mortality !Tlay be of great significance in the popUlation dynamics of the host, but it need not necessarily be the case). It should be added here that I have not made a conscious effort to locate egg parasites, and some undoubtedly exist. However, in collecting pod samples in the field, egg parasites would normally be picked up along with the rest. As Hill be discussed at the end of this section, lawaI' zero levels of parasitization of tropical insects may be cOlT,:nonp::'ace; t;-Je follc-.ring paragraphs are directed at why they are so low among Guanacaste Bruchidae (there is no hint of such a phenomen'.)n arr,ong ;;orth Americc.n Bruchidae) .
The absence of hyperparasitcids is associa:e:::' i~ one case with the ?resence of an e:f::ective disease. T:--:e :-.. ;:.' ~r-'-=:::·:.~'- ~ :.L :;::~;~:?e:'2~
::i.€ in a sa:Tlp=-~ c~ nu::s. 7:-1'2SE: ceac ~~':'.::.:--::'C-:,~=- "-'~.'~"_
free lar':ae of the weevils in Andira en8rmis ses::s ,,:::'-Ie a~ 'O:~~e::,-i ',''0:
fungus disease) and this bruchid's habiTat is c~e weccE:S7. c~ :::.l: bruchid microhabiTats. I assum~ That t~e eas~:7 acce~s~ble 5c;zeeZea nut bruchids (they are in the n~ts for 2 to 1: ~~~ths) are LaT attacked by a hyrr;enopteran hyperparasitoid beed'lse :rJ';;-: Houle be ei ther "esten" by the disease or indirectly o:.:'_corr,;:Je:ej by .-1 t Hould seem virtually irr,possible for the h:'7.e;-0iJterar, to oc.::co;;;!=-'et'2
178 DANiEl H. JANSEN
tr.e b2cterium yet allow the br'Jchis lr"I'va ~') live lDrl;~ C';L')UGr. -:') c~t its exit hole out of the hard palffi nut.
I turn to properties of the bruchid-seed interaction to explain tte other cases of low or zero hyperparasitization of br~chids. Some thoroughly examined cases are the following. Cassia grar.dis has three species of bruch.~ds in' i ~eeds, none of which have hyperparasites in Guanacaste (though. the e~g clusters of Pygiopachymerus iineola are very~arely attacked by an unknown hymenopteran, Jc.nzen 1971d). Pithecellobiwn sama:n has one bruchid (Merobl'uchu3 colU'nbinus) that kills 40 to 80% of its irr.mense seed crops (a large tree may produce 150,000 seeds annually) and no parasites have jeen reared from 85 samples of 10::l-plus pods taken over a three year period. The bruch~~s'have been reare~ out of 23,285 fruits of Guazuma ulmifolia (samples from 205 trees in six major habitats) and the two species of bruchids infesting as high as 99% of these seed cr~ps had no hyperparasites (Janzen 1975b). Of thol:.sands of An,blycel'us bruchids reared from Cassia emarginata pods, there rave been only two large Chalcidae, of a species customarily reared from a wide variety of Lepidoptera pupae. A sample of better than 20,000 Mimosestes reared from Caesalpinia cor·iaria pods had no r:yperparasitoids.
Examples such as these are probably explained by tte interactio~ of severc:,l.·factors. To be a bruchid hyperparasitoid, the hyme:-.':l}> teran must n.o:t qn~y- be able to survive in the bruchid larva (l-;::ich may b~ complicated by the presence of secondary compouncs in t~e larva) but it must be able to get to the larva. However, t~e ~~re
species 6f b~~chids there are in the habitat, the more s;:ecies r": bruc~ids the hymenopteran will have to be able to attac~ in O~:~~ to be able. to .. a"Ccum-11ate enough hosts to stay in the ["':-:."0. ;~.e ~.es:
for this increased generality is predicted because the ~c~e s~~~~ss of bruchids ther~.are in the habitat, the fewer indivi1~~ls t~"'~s
are on the ~av~2-~?J~e in e~ch O:-1e of then,. HOHev.-::r., v';:;!'"'::':;'-::: =-.: --=",: :l2.!-ficul t in the ~e;t 0:-: ~:7":.~s. r.S tr . .:; ~"~"~C11': =:; '::'1~'-:: ~~ ~ - -". - r
?eneral case. ASSLi~ie a [rc:~ie:-(t :=rO:1i 0;--J-:'; very CCi~.~~;=L ;,r __ =--: s;:;:~:.~~::,
to a m::>derate nUil:::,er of r.jC):J.era-rely COTi'::;,un host s~ecies ~0 a "'';S-:··
large ml~ber of quite scarce host speci"s. Likewise (j::::::-,~.e t:.~-:: -::~.<::
total bic:T.ass of bOSTS does not cnan?e uler this gracL<:::-,-::. ~;.'"
nu:n~er of parasitoids t:1at cC.n be supported along this prc.c.;.e:-.:
INTERACTIONS OF SEEDS AND THEIR INSECT PREDATORS/PARASITOIDS
expected to rise -'';'rst toward the middle, and then at the limit falloff to a point considerably lcwer even than it start~d out.
179
r view the moderate numbers of bruchid species to be found in southwestern North America, each with several species of hyperparasi~es, as representing the peak in this curve. The more dis".:incti?;e is the average host species along this gradient, the more rapidl~'"'"aI'ld severely I expect the numbers of parasitoids to decline at the upper levels of host sp~:ies richness. Likewise. the more fluctuating (predictably anq ur.?redictably) the physical environment along this gradient, the more rapidly and severely r expect the numbers of parasitoids to decline at the upper levels uf host species rlcnness. Thus, for example, I expect the deciduous tropical. :tor.est with n (large numbers) species of hosts to have as few species of parasitoids as an adjacent evergreen tropical forest with 2n species of hosts, even if the total annual production of host biomass was the same in both sites.
Following this type of reasoning, I expect a number of Guanacaste bruchids to be very corrUllon (in most years) yet have no hyperparasitoids, simply because on some years the prey density falls cO very close to zero. In short, I am saying that the more tbe availability of a lower member of the food chain fluctuates the density, the fewer specialists can survive at higher levels in that food chain. As a case in point, there was a general drought 07er much of Guanacaste.at the beginning of the rainy season in 1971. In many areas it was sufficiently severe to cause the abortion of almost all pod crops of Pithecellobiwn samano How !-feY'oDY'ucizus colu'nr..inus, -che host-specific seed predator of P. saman, managed to survive is nOT clear (it probably re-immigrated from neighboring areas of less severe drought), but I can easily envisaEe that a host-s?e:ific hyperparasitoid would have had an even ffiore difficult ti~e surviving.
. -an~ ~. -~:e~: ::a_:~::.
-che 12~vae nor "the ;:L;2e sL~J~: cn)~ .si:~~
~hou£Zrl at S8~.e tr~es ':c.:::-'s-::- :-.-,_:--.j~:r'~; d~~ r-:
the newly e;ne::-'ged ac. __ llts GO noc rETurn to 0\-iposi: or, :;-,LC :.':0',; c.::--_~
of foliage produced by tt€: j",,::"oliated or ca::,2gec. CY'e"". ;'. -,e:-,--:a: >12 hypothesis is suggEsted by trlese obSer"J2: i~ns. .It ITlaV ~~--: -_~.-:;.:- -'::.:::
herbivore is genetically ?::-'ogram:ned for a time when its~.-,-::-cs -,r""re IT.uch harder to find (rather chan ~eing e:-:;)osed tr""es por:i~.;; 0;.;.t of pastures as is currently the case), and cOTal c.efolia-c~~~ a much rarer event. -,:hen the "';.r""e is und""y heavy cc,::.?eti :is:-_ i r. ::--'" forest, iT produces one larze new leaf ero? (the une t~a: __ :e larv2~
180 DANIEL H. JANSEN
feec on), and then turns off new leaf product i,':1 eo/en when a few llave Leen removed by herbivores. In the fores:, the herbivore gets one shot at the tree, builds up a moderate population of adults, and stay~ in the game by being an active adult in reproductive diapause for eleven months until the next new set of leaves appears. Such behavior appears to be adapti~e in the context of the individual female, in that she may ge~ more'ev~ual offspring by waiting to oviposit rather than by exposing h~self to predators while searching for a few new le~ves, and by not producing caterpillars that Hould be members of.~ relatively slT,all overall body of arthropod prey items which would be confronted with an array of predators developed on the burst of prey produced by the big flush of leaves at tte beginning of th~ rainy season.
The hyperparasitoids that do survive on Guanacaste deciduous forest bruchids are worthy of much more scrutiny, but this portion of the stud~ is still in an e~bryo~ic stage. Inspection of the material reared to date along with the bruchids reveals one conspicuous fact. In strong contrast to the highly host-specific bruchids, a given species of bruchid hyperparasitoid is found in many species of bruchids (and I suspect, in many species of other insects inside of seed pods and similar structures). It is of interest to note he~e that it is the rare bruchids that should be most seriously affected by such Hymenoptera. For example, the hymencpteran.might be presented with an array of 1110 pods per hectare, which might for example contain 1000, 100, and 10 pods of three similar legume species each of which is attacked by a specific bruchid. If it fjnds 30% of the pods, a 30% reduction in absolute numbers of the rare bruchid is much more likely to be lethal to the l?opulation than a s.imilar percent reduction in the cO;;-.IT;on bruchid.
As "c,lentioned at the beginning of this section, tr.e finding thaT a number of bruchids ~ave no hyperparasitoids and that only a few sustain severai E....~ej:ies of hymenopterans is in a;r",e::-ien: I-I"i th my g9neral find~n?s ~i~t) s~eep s~mples of the ~~S~2 F.ica~ ~~:hrc'~oj
r~chness to t:-;.;:; ~Dgl':";:,;; :::eac.:'J'''; ~Tas rainfores: l';':Jjsrs:-:.~'y ~.:~ ::- .. t lyJi:::12 as m~ny species of prey insecTs (Janzen and ?~nd 1975). Si~"ilar
A:1des; Hitfl increasin" elevetion, the hymeLo~"!:eY'on ]:-ardsitoic species richness was proportio~otelj the least reduce~ G~ o~~ ~rouDs, pres~~ably assoc.id.tE:(~ v.?itb t~.-2 f2·:.t that -rIle nUifILr.=y,;:... c~ ~Jr-=y: "~:l~i·.".jduals
INTERACTIONS OF SEEDS AND THE·~ INSECT PREDATORS/PARASITOIDS
per species was highest c' the highest elevation (J;:jJ.zr"n et 21.
1975) .
SUR'/IVIllG THE INIMICAL SEASON
181
.In the li~ of our ;r,id-latitude biases, bruchids (and other seed parasitoias) in 'the G~anacaste deciducus forest do a very
-:'peculiar thing. Upon corr,;::.leting larval development, they pupate and emerge within a few weeks. In rnost species of hosts, there are not adequate numbers of seeds of the a~propriate developmental stage for them to produce a second generation. The adults are :hen ac.tive, but reproductively inactive, for 9 to 11 mon:hs until the ne~t-'seed crop appears. Sorne feed on the necta~ and pollen of their host pla'Jt's flm-lers (e.g., Acanthoscelides pigY'ae on Mimosa pigY'a described earlier; Acanthoscelides oblongoguttatus taYes nectar from the extra-floral nectaries of Acacia cOY'nigeY'a, its host in Veracruz, Mexico, Janzen 1~67), or on nectar and pollen of other species of flowers. However, most species may be ta~en by general collecting with a sweep net in vegetation that is neitter rich in flowers nor contains the beetle's host plants. Presu::"'1tly t[;e adults have "-' lower mortality rate if they actively s,,:d:: out appropriately moist, cool, or shady microsites, and if th,,:y can actively avoid predators, than if they attempt to survive as a d0rmant individual in the damaged seed or a pupal cell in the ~round or
. litter. In this context, it is of interest that the only Guanacaste bruchids with a cons?icuous mortality from a disease are those that wait for long tiT:les i;-, palm nuts on moist litH:::::-' :.oe:=ore er:lerging. Furthermore, tr.e only seed parasitoid wi t~, a cons}:i-::'.1ous fungal mortality are the r.-l::> Cleogonus weevils that c-: -:aCK ;"rv:iY':.z
eneY'mis seeds; the larvae ;".1~ate in the moist s::>il t,ec. '"'.-: thEe 'SEeCS
and may lose better than 3::"0 of the popUlation to an ,~~;-~esc:r~:.,=-:
species of fungus. In s~.'Jr·t, I am saying that the se"'s :,o,Ja=- ~ ::·ias are not exceptions to the re~eral hypothesis (.·a~ze~ :~7~:)
a ]2r,;z-e ;-;~:-,:'2r of trc)}-.:.c;~~ ':':-.:02::-:: ~·a?~ :he .:;-.. ~-.~:':::G~ - (."' r.
0"f.>c:::.ec ("~~-\ic::<=- ::r~,.:i:s 0:= 2-::,",,: ~":·2::,::Y~.::2 (L:i;:2C:C~.'2): "'c~~~n C~~:e?-iT1g of a fcTi:yr ~:.;_ -~r~:::-,~c;.,:" o~~ }-~d·re Si/l:'O~ ;::: ':Y',"'./~ .... ~t:'-- s' trj~ir h:):=;:-sr)ecific seed -:~:::::,=-:::rJY- Ths ~~ar2J2 C~lJS:,::~= r:- c.r",' v:iLG-di~.;:''2rsed hZvar-aao2 G.:--r/Y'?tz(;i-::es (Sinaro1.1L~j':'2Gs) :-?' ~~-::s :j;~~:~ besD founc TO C9Dl:ain 2r2\"'='" -_s o~ ZQbr07;e~ sp., Seny:iv.::: ~--:J!-'ccU!..r~ 'l
Sel,nius sp., two species 'J:= ,;canthoscelic?es, Pm'a;7G?";.,,: <;~.d phiZ!1:des (Curcul':or;idae); h. a}noi7p~,:;icieE b~s llO pre-dis~~rsal =t::..;r:;,: wY'~~-=-:')rs that live in the seeds. ~imilar arrays of adult br~~~:~: hav~ ~Een
182 DANIEL H. JANSEN
found in the dry infructescences of TY'iplaY'i[ ameY'1..cana (Polygonaceae), ano~her tree with no pr~-dispersal se~j predators in its seeds. All three of these records are from the last half o~ the dry season. A sweep sample of a stand of pure BaltimoY'a Y'ecta (Compositae) in flower in the middle of the rainy season yielded ten males and eight females of CaY'yedi;.s quadl'idens, a bruchid that breeds in the dry season pods 01;- Centl'osGl"'I6I:...plwnieY'i, a legume vine. S ... :eeping in nearby more mixed "o::..d field" vegetation yielded CaY'yedes quadl'idens, Amblyc~~s cluvnpioni, Sennius instabilis, and two species of Acanthoscelides; the seed hosts of none of these bruchids were . available at that time (mid-July). In the adjacent deciduous forest unde~story, simultaneous sweeps produced adults of CaY'yedes quadl'idens, C. cavatus~ C. :p-7-itw'us, Ctenocolwn tube2'culatwn, tW:l species of Acanthosce lides, alTd --a species each ·of Amblycerrus, ZabY'otes and" Dahlihl'uchus. With the exception of the last species, whose host is unknown, the seed hosts of none o~ these bruchids are available for ovipositio~ at that time, and only the AmblyceY'us will eventually find its host (Spondias momb.in seeds) within a few meters of the vegetation swept. In the middle of the dry season (March) a sweep sample primarily underneath scattered evergreen trees in the sar-.e forest site yielded the following adult bruchids: CaY'yedes quadY'idens, MeY'obY'uchus colwnbinus, M. solitaY'ius, GibbobY'uchus gUGl74-
caste, Ctenocolwn zubeY'culatwn, MegaceT'us impigeT' group, two spe~ies of ZabT'otes, three species of StatoY', two species of AmblyceY'us, Acanthoscelides quadY'identatus, A. megacoT'nis, A. peY'tinax group, A. n~. bY'ev.ipes, A • . sp., Sennius instabi lis, Sennius moT'OSUS, Mimosestes sallaei and M. sp.; the seed hosts of all of these species are. well into their infestation cycle by this time, and oviposition has occurred long ago for most species. These adults
. were undoubtedly newly emerged adults beginning their long wait until next yeaTts' seed crop. It is noteworthy that one of the most common large bruchids in this forest rich in trees a"" 'Juaz~c"'::;:
ulmifolia is ~ts bos_t-specific AmblyceY'us cistelinus. Hh'le la,?e nunbers of these.-beetles Here er"erl':ing from (;. uZ-rr,ifolia fr'Ji ts near ~tle si:e of t~Je c~\r ssassn ~n~er'st0ry s~~e:. s~mples, G : c:.~
c~ these L~etles ...... -=.2 ~G.~~-2I' :~.-:':::''2.
T'" -.-~_ •• _ - • " -,~
.Jl~~ _':--=:-,.,~.:..
seeds an~ 'thEir ;.:i::2si.-::.c.i::::.~ ~n L2~::--=-a-:S -:_~=: L2~/~ :.2'2r .. ;,-:;rtuY:'':::-/_ in various ways expl~i~a~ive ~~~~ern 6zrjcul:~re and technols~-ov~r ~he last 100 to 380 years (and see Janzen 1373d, 1974c). The probleffi is raTher s:raightforward. ~hen a set of species evolve with respect to each other for thoLsands of generations, and part are then removed and the reffiainder have their densities and othe:r- properties (e.f;', timi,jc of ne'v.' leaf pY'oduction) alte:--,,::::, it becomes extremely difficult to interpret the adaptive signific~~ce
INTERACTIONS OF SEEDS AND THEIR INSECT PREDATORS/PARASITOIDS 183
o"f many c)f the charact·.-:---istics ger.t:;ticall·/ prograrr.ijied Int,=-, -_~.c.;
individuals. Of course the species that are there are still in~eracting w~th each other, and therefore it is quite possible t~ examine their contemporary ecology, espe2ially from a mana~e~en~ viewpoint. Howev~r, the generation of e201ogical principles, to say nothing of ~evolutionary ones, is almost impossible when \olcryir,g with sy..stems thcI~re not only not at equilibrium, but have b<.:en rec:",~,tly pushed off "equilibriwTi to an unknown degree by unknmm per't'.lr'~aT i0;)S. for example, the pastures of Guanacaste are full of swoll",n-thorn acacias containing' three species of obligate acacia-ants (PseudcmyY'mex belti, P. fel'Y'ugir;ea and P. nigY'ocincta), all fh;hting a2c:ively over and coexisting on a single very narrowly defined reS,)1..:r2e; how can one understand species-packing in such a situati0D when all the physical'en~i~onmental barriers that probably were once q1..:ite important in separating these three species have been obliterated to provide the United States with hamburger?
The bruchid-host interaction is particularly suscept:~~e to the thorough types of habitat and species destruction practiced ~y modern man. In trying to understand how the sizes and c:i~ing cf seed crops may have been influenced by the interaction wi~h the bruchid, I need to kriow how long the seeds of a given crc~ are available to the bruchid. Except with wind-dispersed s~ecies ~his is no longer possible owing to the effective elimination cf all :~,e
large dispersal agents. Even where dispersal agents are ~ :-~3"":-;t, the numbers of seeds they remove is closely related to v:;,at st:-,e:food sources are available, and contemporary deciduous forests ~~:t their numerous edges, and old and new fields, are a vas::~ ~~~~~:-s~t food resource base than a relatively undisturbed deciduo:..<s £'0r""s: with scattered indigenous farming efforts. furthermore, cs c~:-:;;;i:-,
dispersal agents are re~oved, the less desirable plants ~~l: s~~~sr disproportionately to the ones with the more highly desir~~ ~r~i:s or 'seeds. Rhinochenus tY'ansveY'sali:: and R. stigma (Cur'-::'~~:-::<-::~",,) used to --ill as mJch 2S soSe ( Cc:esal: ()CC1Jr'S i:. I.l,:jS:: ~~'. CCX)1-·~'::";'l:'t r·Dt!'C':"~""=::·~n:::. : .• 2t~,:":'~~ ~:::: --
bE:~>~1...:.se it c.e?E:12S 0:1 :'is~)e~sa} ~'--'0~.-:S .:~\ '--:->::~ 1r-.~ i:-1=-=·.:'~~~_-r,
so it caT) e~,s:r2e (.-:f. \.-CY:Z'2:-1 lCj~7,-=. S:.~~ 2-;- i,~le i:-. ~.'" - r'
ShOl11d be s~ressed ~ta~ ~:atitat 21-~~a~j()~ ~y man ~s n~
(e.p., Long and ~:2rtin :07L ), ~u+ ~~ is ~0~: ~CCU~~~~? ~~
at ~ rate and in~ens~~v ~ar surp~2~~Lg ~,r~~~o~s ~~~~~S.
I also need to know ~.~~ many jruchi~ feT~ales =a~ ~~ S~::.~ __ ~_ to arrive at the host plan: seed crop. 2y ~lteri~~ ve2~~~~~~~' ~
and changing the ~elative abunda~ce of al:er'~ate aj~lt ~~~~ :~_~~~:~
cor.temporary agriculture undoubtedly increases the nu;;,:'.er:, ,,;r_·:_~.:
at some plants and species. and dec~eases those arrivin? ~: s::,~~=.
A Scheelea rostY'ata paID growing ir. an ODen pasture wil: ~~~e seed crop almost enTirely free fro;:-; br-uch.i~ attacK., · ... :r1~.::.J:.: -:':-.~ ?-:-"'~·t::-:-.::
184 DANiel H, JANSEN
20 m away in forest may lose 80% of its seE~S (:anzen 1971b). lio'.le'fer, ScheeZea rostrate: adults left grov':'r.g in pastures w[,en the forest is cut are effectively seed-sterile, since the dense seedling shadow i~ eliminated by the fierce dry season sun. Introduced species from other parts of Central America may be expected to maY-e this part of the story even more confusing. H. Dingle has even suggested that the Dysdercus ·th;;t.t....::ills s9 many seeds of Sterculia apetala in G~anacaste (Janzen 1~72) may be an introduced species.
-'
The effect ot all of this is, of course, to cause one to seek out somewhat less disturbed areas for study, such as Santa Rosa !Jaticnal Park in northern Guanacaste and the forest near 'the OTS Palo Verde Field Station (COMELCO Ranch) between Bagaces and the Gulf of Nicoya. Ho~'€ver, even these .£i,tes are ephemeral in the f face of irrigation schemes, inflation, population pressures, and private interests of surrcunding ranchers. The story is a familiar one, but
4ecologists have a tendency to think of themselves as racing
cgainst time. From what I see around me in the tropics, the race is already lost.
ACKIWKLEDGEI1EJ.:TS
This study has been supported by NSF GB-25189, GB-78l9, and GB-35032X, by the Organization for Tropical Studies, by the Universities .o.f lSa!,sas, Chicago, and 11ichigan, and by the intellectual and physical input of a very large number of people Vlorki~g for and with these institutions. Thev are mentioned by name i.n the spec ifi c pape'rs dealing v:i th their co;tribut ions. The study \o!suld r.ot have been possible without the taxonomic studies of bruchids an~ weevils. by. John M. Kingsolver and D. R. Whitehead. J. A. Chemsa~,
~, D. Duckworth, S. L. Wood, R. ~arner, P. A. O?ler, P. Wunde~lin, \\'. C. burger, and .i?-bove all V. E. Rudd have con'tributei to de:er-rr:i r~a t ions or o_1;"p.er groups. I.. E.. Liener ar:.Q E.. n... Bel: contr':::'·~ ted IT'~r:y of the seco:-l':::'~r\~ cc;~~,··::)unds :..:.se~ ':T; ""':".:-,~ :-::€:::~;-l~ t.::.;.=-::s. ~'"=:-.-
':':E::-~ter', :. ~."I a:l~ c. ~I. J=.hLs'~ri. 1;;;7LJ.. Cc,€:·,is.1.l..o.Lion c:- ~)':Jr..~ =_-=c.:;-: beetles (Cole~ptera: Eruchidae) and their hssts, E~ology 55:10S6-1103.
[vans, R. J" aD::! S. L. r.ardemer. 1967. nutri:tive vo::"\.:.e of .:.e;::',,;:n€:: seed proteins. J. tgric. Food Ch€::m, 15:43S-443.
Gwynne, ~,D. IS59. The nutritive va11.108 of haacia ~~ss lD re::"a:ion to A~aciQ se>:=2. dist!'itut.ion by ungulatl?s. East A~r. Hilc.l. J.
INTERACTIONS OF SEEDS AND THEIR INSECT PREDATORS/PARASITOIDS
7:176-178. Janzen, D. H. 19l5. Coevolution of mu<:ualism between ants arId
acacias in Central America. Evo11lt ion 20: 249-27:'.
185
Janzen, D. H. 1967. Interaction 0:: the bull's horn acacia (Acacia cornigera L.) with an ant inhabitant (Pseudo:r;YY'17JeX ferruginea r:;:Smith) in eastern !-lexica. U. i<ans. Sci. Bull. 47:315-558.
Janzen,i.~ H. 1969. Seed-eaters versus seed size, numLer, toxicity ami dispersal. "Evolution 23: 1-27.
Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. Amer. l:atur. 104: 501-528.
Janzen, D. H. 1971a. Seed predation by animals. Annu. Rev. Ecol. Syst. 2:465-492.
Janzen, D. H. 1971b. theparen1: tree: 15: 89-101.
The fate of ScheeLea rosta>ata fruit:s teneath predispersal attack by bruchids. Principes
Janzen, D. H. 1971c. Escape of juvenile DiocLea m~gacay>pa (Leguminosae) vines from predators in a deciduous tropical fcr~st. Amer. t:atur. 105:97-112.
Janzen, D. H. 1971d. Escape of Cassia grandis L. Leans from predators in time and space. Ecology 52:964-979.
Janzen, D. H. 1972. Escape in s?ace by StercuLia apetaLa see::s :=rom the bug DysdeY'cus fasciatus in a Costa Rican deciducc;s forest. Ecology 53:350-361.
Janzen, D. H. 1973a. Community structure of seconcary cor..pounds in plants. Pure Appl. Chern. 34:529-538.
Janzen, D. H. 1973b. Comments on host-specificity of tropical herbivores and its relevance to species richness. p. 201-211. In V. H. Heywood, ed. TaXOnOl;]y and ecology. 5yst. f..ssoc. Special Vol. No.5. Academic Press, London.
Janzen, D. H. 1973c. Sweep samples of tropical folia;;e insects: effects of seasons, vegetation types, elevation, time of day. and insularity. ECOlogy 54:S87-702.
Janzen, D. H. 1973d. Tropical arroecos~st~~s. :te~e ta~~~ats
are misunderstooc t~.'" t;.s te;~,~ t=:rat? zr.;Les, ::~':;"~:,:',~:-jr~;'~ -'" ..,.
i-iist. 83:48-53. Janzen. D. H. 1975a. Sweep sa~p1es of :ro~ica: de=idu~u~ ~or~s:
foliage-inhabiTing insec~s: seasonal changes and Int~r-fi~ld difference~ in adult bugs and beetles. [colo~1. In press.
Janzen, D. H. 1975b. In~ra- and inter-ha~itat varia~io~ in G~~~u~a uLmifolia (Sterculiaceae) s"''''~ predac ion by i.'?~bLyc8r"-s cictelinus (Bruchidae) in Costa Pi::a. ECOlogy. 1:-1 py·ess.
186 DANIEL H. JANSEN
.Janzen, D. H., 11. Atarroff, H. Farinas, S. RE:."'S, 11. P.inc:on, P •. Soler, P. Soriano, and M. Vera. 1975. C~anges in the arthropod community along an elevational transect'in the Venezuelan Andes. In press.
~anzen, L. H., and C. M. Pond. 1975. A ~omparison, by sweep scmpling, of the arthropod -fauna of secondary v<;;:getation in Michigan, England an9 Costa~. Trans. R. EntoIT,ol. ~,oc. Lond. In press .
.Johnsen, C. D. 197J: A new Acanthoscelides from Indigofera (Coleoptera: Bruchidae). 201. Bull.' 27:169-174.
Johnson, R. M., and W. D. Raymond. 19~4. The chemical co~~ositi0n of some tropical food plants: II. Pigeon peas and Cc>W F-eas. Trop. Sci. 6:68-73.
Liener~ I. E .. 1974. :PhY1:ohaerr.agglutin~ns: their nutritiorJal f significance. J. Agr. Food Chern. 22:17-22.
Long, A., and P. S. Martin. 1974. Death of Ame~ican ground sloths. Science 186:638-640.
1':agnus, K.l.~and C. E. Sea~orth. 1955. SarnaY'.ea sarnan !':errill: the rain tree. A review. Trap. Sci. 7:6-11.
Rehr, S. S., P. P. Feeny, and D. H. Janzen. 1973a. Chemical defence in Central American non-ant-acacias. J. Anim. Ecol. 42:405-4 16.
Rehr, S. S., D. H. Janzen, and P. P. reeny. 1973t. L-vopa in legume seeds: • a chemical barrier to insect attack. Science 181:81-82.
Rehr, S. S., E. A. Bell, D. H. Janzen, and P. P. Feeny. 1973c. Insect i'~idal a.m~no acids in legume seeds. Biochem. Syst. 1:63-67.
Rosenthal, G. A. 1972. Investigations of canavanin~ ~~0che~.~stry in the jack bean plant, Canavalia ensifOY'r:;is (:'.) Canavanine biosyn~hesis in the de~eicpi~g plan~. 5): 328-331. ...
DC. .!. 1 .
~;evilla-E~sebio, C., J. C. M6~,~ril, J. A. r~s~~ic, ~;.~ ? ~. ~~nz~~s~.
