Zoological Journal ofthe Linnean Society (1981), 71: 265-277. With 3 figures
Food sources and foraging tactics in
tropical rain pools
A. J. MCLACHLAN
Zoology Department, University of Newcastle ujon Tyne, England
Pools on exposed rocks are coininon over much of Africa. Based on dimensions and position, those
exaiiiinrd arc of' three types. Each type is inhabited by larvae of virtually a single dipteran species at
Iiigli drnsities (over two million larvae in-?).
Location of the pools suggests that food might be a limiting factor. However, events, including
defecation in pools by civets and genets, fruit fall and wind-bornepollen, apparentlyensure that this is
not the case. In this environment of superabundance animals are presumably free to choose favoured
items of food.
Each aninial species does, indeed, take a characteristic assemblage of food items. However, each
species is shown to eat whatever it can swallow, differences in gut contents being due to differences in
tlic clratactctisric food iterns available in each type of pool. Most algae are excluded because they are
too largc o r iiiaccessible, which means that the pool food chains are based largely on allochthonous
detritus. There is no reason to believe that food type, perse, has any influence in determining which of
the three dipteran species is present.
KEY WORDS:-
Rain pools
- Tropical Africa - Chironomus - Dhyhelea - Pohpedilum.
CONTENTS
.
. .
Rationak
.
Fieldwork .
.
.
.
.
Experiincnts
.
Lalmratory procedures
.
Rrsu I t s
. . . . . . .
Types ofrock pool
. .
Inrtotluction
Mrtliods
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Relationship between pool size and dipteran larvae
Location ofpools
. . . . . . . . . .
Densitiesofdipteran larvae
. . . . . . .
Algae and allochthonous organic matter . . . .
Nutritional characteristics of the pools
. . . .
. . . . . .
F o o d and feeding of dipteran larvae
Discussion
. . . . . . . . . . . . . . .
. . . . . . . . . . . ,
Ackiiowletlgelneiits
Rcl&wices
. . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
,
. , , , .
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . . . . .
,
.
265
26 7
267
268
268
269
270
270
270
210
270
211
271
272
214
276
211
INTRODUCTION
Temporary waters are characteristic of the tropics and subtropics with their
sporadic rains, high temperatures and associated high evaporation rates. Lasting
for anything from a few hours to several years at a time, such waters range in size
265
0024-4082/81/030265
+ 13/$02.00/0
0 198 1 The Linnean Society of London
A. J. MrLACHLAN
266
I . A . Kairi pooh o n ruck surlares in tropical Africa. B. Granite 'whale-backs' scrn in the
middle distance, o n which the rain pools occur.
Fig1ii.c
TROPICAL RAIN POOLS
267
li-om lakes like Chilwa in Central Africa (c. 1000 km2 in area) to puddles in tree
Iiollows, big game footprints and depressions on rock surfaces, all holding only a
few ml of water. Some of the types of rain pool encountered in Africa are
described by Rzoska ( 196 1) and McLachlan ( 1974). Similar situations exist in
Australia (Williams. 1975; Jones, 1975) and South America (Junk, 1970; Reiss,
1976). Although less typical, they occur in temperate regions as well and have
attracted a limited amount of attention (Moor, 1970; Hartland-Rowe, 1972;
Disriey, 1974).
Because they are often sinall and tend to occur in remote regions, temporary
waters have been overlooked and in a sense they are, along with large rivers, one
of the last limnological frontiers. In addition temporary waters can be of
considerable economic importance as a source of rice and fish. The ecological investigation of such situations benefits because these small ecosystems are
unusually amenable to study. There are several reasons for this, among them is the
fact that, since few organisms are adapted to conditions of continual wetting and
drying, temporary pools are relatively simple systems, a situation apprec‘iated by
Williams ( 1972) working on Australian temporary and saline waters.
My interest is in tropical African rain pools like those shown in Fig. 1A. They
vary in size froni about 100 cm to less than 2 cm in diameter and are virtually
unknown apart froin the work done by Hinton (1968) on the mechanisms of
drought resistance of one inhabitant. Pools usually occur on large ‘whale-backs’
which are often several square kilometres in area and several hundred metres
Iiigh (Fig. IB). Those shown are in hollows resulting from natural exfoliation
processes. They are also found in solution hollows or in artificially created
depressions such as those made for the placing of explosives or as a result of
gr-aiii-pounding activities. Several pools occur on a single whale-back so that
o1)scrvationsand experimental manipulations can be replicated.
111 SOIIIC cases entire communities, including arthropods, can withstand
tlcsicc-atioii and it is therefore possible to dry the pool contents and to transport
;in allnost intact ecosystem for storage and study in the laboratory.
Some o f the biological characteristics of African rock pools from both north
;ind soutli of’the equator are considered with special emphasis on the feeding of
the iiiscct larvae which form a major part of the animal communities.
METHODS
Rationale
Preliminary observations in the field led me to believe that several distinct types
ot’ pool exist. Large pools appeared to be associated with different aquatic
ai.thiupods and to receive different kinds of allochthonous organic matter from
thosc in siiiall rock pools. Assuming that the conspicuous input of extraneous
organic debris is important as food for the animals, the following features
were examined : pool dimensions ; macro-invertebrate densities and species
win p osition; and macro-invertebrate gut contents. Kinds of organic particles
occurriiig in the mud and water were examined. Special attention was given to
the occurrence and type of allochthonous organic matter and to feeding
bchaviour 01. insect larvae inhabiting the pools.
_ ’
268
A. J . MrLACHLAN
Fieldwork
o n l y simple equipment was needed: a metre rule, a 3.3 cm diameter tube for
collecting rnud cores, ‘polythene’ bags for storing core samples, 100 ml capacity
tlwnc’ bottles for water and surface mud samples, formaldehyde and
ibi-e fi1tc.t-s ( 5 ctn diameter ‘Sartorius’ SM 13400). The latter were used in
ing dry samples of particulate material o n the mud and in the water for
The filtering apparatus consisted of a ‘Sartorius’ filter head
le pump with reversed piston and one-way football valve, as a
vacuulti piitiip. l h e length, breadth, depth to the highest water mark and mud
depi 11 \\ere iiic;isured. Water and surface tnud samples, normally collected with a
Pastcur- pijmte, \\.ere passed through the glassfibre filter discs until the filters
~ - C I XI>u l i y loaded. Filters were immediately sun-dried and stored in a desiccator
ILr c-lieii;ic;tl analysis. At the same time similar mud and water samples were
collected arid preserved in 4%formaldehyde for later microscopic examinatioti.
To assess macro-invertebrate densities, triplicate mud cores were collected
horn each pool, mixed and preserved in a 4%formaldehyde in ‘polythene’ bags.
Practical considerations prohibited the treatment of the three mud cores from
each pool separately. Despite the resulting loss of information on variances,
lumped means from several pools give, in this case, a relatively precise estimate of
population density. This is partly because the mean arrived at with three samples
will obviously be a better estimate than that derived from one core, even if
variances are not known. However, the peculiar nature of the pool environment is
itsclf relevant. Among freshwater habitats it must be uniquely easy to sample.
Apart fi-omsmall size and shallow depth there is a tendency for a uniform layer of
setlinients to form. Many pool inhabitants build U-shaped tubes in this mud. The
position ofeach tube is visible to the observer because of a chimney attached to one
end. Such tubes can occur at unprecedented densities and show that close packed
individuals are uniformly distributed over the entire bottom of the pool.
Another point associated with the small size of the pools is that even three
3 . 3 C I I I cliatiietc~rmud cores remove a substantial proportion of’ the habitat.
Delihit\- estiiiiatcs therefore approach population parameters rather than
p o p 1 ; r i i o i i statistics. This is clearly desirable except that the population itself
niiglit be timiaged bv the sampling procedure. However, only one set of cores is
takc.11 pci. 1)ool a t i c t animals are in any case continually being replaced by
o \ . i 1 ) 0 4 t i o t i . While the pools are floodect during the brief rainy season casual
iiispec,tioti reveals, not surprisinglv, no gross variations in population density.
h r o t a 1 01’ lif‘tv pools \<ere exarnined, forty in Malawi in the vicinity of Lakes
Cliili\.a ;iii(I Malaivi (previously Nwsa). The retnainder were located in Central
5 i g c . h o i l tlic. sliores o f Lake Kairiji. The full analytical programme was carried
out on 19 of these, all in Malawi.
Experiments
biclti i.csulis suggest that some food items are too big to be eaten. To
iti\,r.stig;ttc. tliis possibilitv i t was necessary to be able to offer major animal
xpcc.ics particles of the same kind, but including a wide range o f different sizes.
Latex ‘microspheres’ (Dow Chemicals) have been used for similar work before
ihtclachlan, Brennan & Wotton, 1978) and are ideally suited to the present in\ ~ siga
[ ti on.
TROPICAL RAIN POOLS
269
After a few days to allow animals to establish themselves in artificial pools in the
laboratory, a few drops of a mixture of microspheres were added to each
container. Twelve hours later larvae were removed and gut contents examined as
described below.
Laboratory procedures
Insect larvae separated from mud cores by sieving were identified, counted and
weighed wet. Dry weights presented in the results were obtained by means of the
conversion factor: wet weight x 0.26. The factor was determined separately from
weighings of 100 final instar larvae of each species.
Food taken by larvae was determined by examination of gut content. For at
least three pools, gut contents of five to ten final instar larvae of the same species
were mixed and sonified to disperse aggregated food at 20 kHz peak amplitude
for 10 s. These were then stained in aqueous erythrosin overnight, counterstained
with IKI and mounted on a 0.45 pm ‘Sartorius’ membrane filter (SM 11306)
which was subsequently mounted moist and cleared following standard
procedures (Cummins, Miller, Smith 8c Fox, 1965).The resulting membranes, at
least three for each species (one per pool or one per container in the laboratory),
were examined using the 40 x objective of a microscope. Particles in at least 20
fields were counted on each membrane giving a total of at least 60 sample units
for each animal species. In order to determine the relative contribution of
various items to the diet of larvae in the field, estimates were made of the total
area ’occupied by particles in each of the following categories: algae, detritus
(dead organic matter and associated bacteria), pollen, fungal spores, rotifers and
rotifer eggs. Inorganic particles, including diatom frustules, were ignored. The
virtual absence of dead algae suggests that what I call detritus comes largely from
the terrestrial environment.
Separate observations were made to investigate the relationship between
particle size and diet of larvae in the field and laboratory. Procedures were the
~ a n i eas those for determining the contribution by various type of particle, except
that the diameter of the first 100 particles in each field was measured.
Chemical analyses of dried samples of mud and water were chosen to give
information on the potential food value of the material, namely protein levels
and calorific values. Since animals are composed largely of protein this seems to
be a useful measure and helps in interpreting calorific contents due largely, in
5onie cases, to refractory material like coal. Protein values were determined by
the Fohn method (Lowry, Rosebough, Farr 8c Randall, 1951) and calorific
content using the Phillipson microbomb calorimeter. Both estimates are
expressed per unit ash-free dry weight after macro-invertebrates had been
removed. Checks against estimates of other authors show that calorific values fall
within the expected range (Cummins & Wuycheck, 197 1: table 3 I1 Bl). Protein
estimates appear to depend largely upon extraction time (Dowgiallo, 1975). My
values are lower than those of some authors, notably of Hynes 8c Kaushik ( 1969).
Analysis of leaf litter identical to that examined by these two authors shows that
m y method gives values consistently lower (by about one fifth). Estimates for the
precision of the analyses give standard deviations of less than 20% of the mean for
calorific values (three test samples of ten replicates each) but rather bigger for
protein determinations ( 17-29% of mean in three test samples of six replicates
each 1.
13
A J
270
VtLACHLAh
RESI’LTS
Types of rock pool
RPlation \hip betrueen pool szze and dipteran Larvae
The occuirente of only one of three species of By larva in any pool, to the
s i i t u d exclusion 01 other species, was immediately apparent. The three animals,
all dipieran larvae, appear‘to be associated with pools of a particular shape and
size (Tclblc I i . Larvae of the midge Chironomus imicola Kieffer occur in significantly
decpei pool\. Student’s t-test, with appropriate modifications for samples of
unequal variance given by Bailey (1959), gives d = 3.4, P < 0.01, F = 14, fifteen
C . zmicola pools u. 32 replicates of other pool types.
Larvae of the other two midge species Dayhelea thompsoni de Meillon and
Polypedtlum vanderplanki Hinton both occupy shallower depressions. While of
effectively the same depth, the latter two pool types are of very different surface
area. P. uanderplanki occupies relatively large pools c. 18 000 cm2 in area
(illustrated in Fig. 1A). DusyheLeu thompsoni on the other hand, lives in pools only
about 1500 cm2 in area. The difference is significant. Student’s t-test with the same
precautions as those applied above gives d = 2.9, P < 0.05, F = 5 . Clearly C. zmicola
larvae occupy the deeper pools while P. uanderplanki and D. thompsoni occur in
shallower depressions, the former with a large, and the latter with a small, surface
area.
1 1
l,oc~/ltiollof pools
I n addition to differences i n size, pools occur in characteristic places. The two
shallow tvpes of pool are normally found at the top of granite ‘whale-backs’
otteti sc.v&al hundred inetres above the trees. The larger C. imicola pools, in
contrast, often occur in old river beds where water has worn relatively large
hollows in the rock. This association normally places C. imicola in riverine forest.
The importance of location in determining the kind and quantity of extraneous
organic matter in pools is discussed later.
Uerisi/w.\ of-dipteranlarvae
Mcan ntirnbcm arid dry weights of larvae in each of the three types of pool arc
also given i n Table 1. Values are exceptionally high, particularly so in the case of
P. vanderplanki, individual samples containing over one million individuals or
50 g clrv weight of lawae per square metre.
Table 1. The dominant species and density of dipteran larvae per m2of pool mud and the
dimensions of the pools in which they occur
1 l i i c . i ~ ~ ) i > oit1
l \ bldldwi. ~ c ~ cin
i iNigei-ia.
Tli~
nunher olpools exairlined ( n ) is given with mean 5 95%C.L. (in parentheses) based, in the case of number and weight
ot lamae. on x
* transformations.
TROPICAL RAIN POOLS
27 1
Table 2. Proportions of algae as percentages of algae plus detrital
particles present in pools as a whole and in the mud and water
taken spearately
Midge present
Pulyptdilutti unndcrplanki
Ua.\yhrlm thotnhoni
C'hirunumu.\ irnicola li
Pool
Water
Mud
N
n
19
(6-32)
37
(2-62)
50
(12-99)
29
(14-32)
11
(6-16)
3
60
50
(2-59)
69
(31-99)
30
(12-52)
30
(12-77)
3
60
6
90
"'Thtec pools in Malawi, three in Nigeria.
Tlir niran and (in parentheses) the range of estimates is presented for each pool type. N ,
Nunibcr ol'pools examined in each case; n, total number of sample units
Algae and allochthonous organic matter
There were gross differences in the kind of dead organic matter (detritus)
occurring in each of the three pool types. As a rule C. imicola pools contain
quantities of fruits and flowers. Dasyhelea thompsoni pools, on the other hand, are
usually almost full of seeds, rodent hair and bones. Finally, P. vanderplanki pools
appear to lack much extraneous organic matter.
At the microscopic level too, allochthonous detritus (and its associated
bacteria and fungi) usually predominate (Table 2). Such algae as there are occur
largely in the water. There are rather few common species, unicellular green
algae predominating in the water and blue-green species on the mud surface
(Table 3).
Nutritional characteristics (the pools
N o differences in calorific value or protein content per unit dry weight of
material were demonstrated between the three pool types, nor were there any
difterences between material in suspension and the sediment (Table 4). The
characteristic allochthonous input, therefore, as well as variations in species and
iiurnber of algal cells are, perhaps surprisingly, not reflected in differences in the
Table 3. Mean and range (in parentheses) as percentages, of four main kinds of algae
present in the water and mud of each type of pool
Algal typr
P. wanderplanhi
Water
Mud
Pool type
D. thompsoni
Water
Mud
C . imicola
Water
Mud
Bluc.-gi-renalgae
Ihlpothnx and Lvngybe
<5
<5
<5
41
(12-82)
0
<20
41
(12-88)
(20
<5
64
(23-100)
< 20
< 20
55
Clirootoccus lurgidus (Kutz.)Mag.
<20
(25-71)
79
(42-97)
< 20
(20
80
138-901
Crcrn algar
Chlattty doinonat
Pvdta~trutnlelras (Ehrenb.) Ralfs.
Other particulars as for Table 2.
< 20
84
(60-100)
0
<5
0
0
A. J . McLACHLAN
272
Table 4.Values for protein(% ash free dry weightland calorific content (cals per g ash free
dry weight) for dried samples of mud and water, determined after removal of insect
larvae
Doiiiinaiit
iiiitlgc.
Nuinl,er
0 1 pool,
P. r~onifrrplmrki
4
I).
8
litorr1~1 I111 2
c. rrrrltol/l
9
Calories 1g dry weight-')
Water
n
Mud
3434
(2962-3979)
3065
12686-34971
3950
(3607-4344)
12
18
15
(XI Protrin
Water
It
3667
(3351-40141
3306
(2783-3924)
3016
(2557-35611
I .6
(0.2-2.9)
0.5
(0-1.5)
1.6
10.2-2.9)
10
24
24
I1
Mud
I1
.5
I .9
(0.9-2.9)
0.6
6
3
(0.1-1.1)
16
5
0.2
(0.1-0.3)
9
Values are means k 95%C. L. (inparentheses).The number of pools examined and the total number of observations for each
pool type, cn)are also indicated. Geometric means and C.L. are presented in the case of calorificvalues.
quality of' organic matter per unit weight of pool contents. Even taking into
i t c w u I i i low cstitnates expected from my protein analysis, both protein and energy
valucs ilrc, if'anything, slightly low for freshwater detritus. The exceptionally dense
atiitnal populations cannot, therefore, be accounted for on the basis of food
quality alotic.
Food and feeding of dipteran larvae
As shown in Table 5 , detritus usually predominates over algae in the guts of
larvae as i i i the environment. However algae are eaten and it is striking that the
proportions in which the different kinds occur in the guts is quite different from
that in the relevant pool. Comparison of Tables 3 and 5 reveals, for example, a
scarcity of green algae in the guts, yet this group forms a large fraction of the
total algal population of the pools. There is a suggestion, therefore, that the
clipterans are selecting some food items in preference to others. The question is
Table 5 . Food of the three dipteran species
Dipteran species
D.thompsont
10
(
C h I ourtic( II
< 10
1-24)
63
(24-99)
37
(1-76)
C. rmicoln
12
13-22)
52
126-91)
27
(9-56)
11
0
(9-27)
0
12
19-59)
R r l a t i w piojxmioiis ol algae as a percentage of algae plus detritu, i n the guts of' t h c
I d n a e a1-r \ h o i v n . The percentage contribution by the four major kinds of algae t o the
rota1 ;tlgac raten is also given. Vdluea are the mean and range. Other particulars as For
Tdl)le 2 .
TROPICAL RAIN POOLS
+
213
I
Lyngybe Tolpthrtx
Chroococcus
Ch/amydomonas
A
I
I
I
Pediostrum
I
Pollen
-__
Detritus
I
C imicolo
P. vanderp/anki
D thompsoni
H
I
m
I
0
I
10
I
20
I
30
I
I
I
I
40
50
60
70
I
80
Particle diameter ( p n )
Figicic 2.Tlic s i m itlianieters) of particles (A) in the pools and (B)in the guts of midge larvae in the
field. Geometric 9596 C.L. (horizontal bars) and particle size ranges (distance between vertical bars) are
givcii. 0111ci.tlciail>a s li)l’Tdbk 2.
whetlier the selection is due to a preference or dictated by other factors. The
latter would seem to be the case.
Figure 2A shows that, leaving aside detritus which includes a very wide range
of’ particle sizes, items like green algae, poorly represented in the guts, are the
larger ones. There is also, of course, an upper limit to the size of particles taken
and Pediastrum) as well
by the larvae. The sizes of the green algae (Chlum~~~monus
as of pollen grains, tend to lie beyond the maximum particle size commonly
eaten (Fig. 2B). This is especially clear in the case of Dasyhelea and slightly less so
in Polypedilum. Finally, it would be anticipated from Fig. 2 that Chironomus larvae
would have the capacity to take at least some of the larger particles from the
environment. This suggestion is confirmed by the presence of substantial
quantities of green algae in their guts (Table 5 ) .
The maximum size of particle taken is related to the size of the larvae;
Chironomus larvae being generally biggest and Dusyheleu the smallest. This can be
deduced from estimates, given in Table 1, of numbers and weights of larvae.
Furthermore, the biggest larvae also have the largest ‘mouths’, measured
between the bases of the mandibles. (Arithmetic mean values and 95%confidence
limits: Chironomus 190.0 f 0.1 pm, Polypedilurn 88.8 rt 0.1 pm and Dusyheleu
88.1 f 0.2 pm. Estimates are based on measurements of ten final instar larvae in
each case.) I t is possible therefore, that no active selection of food items is
iiivolved. Assuming that ‘mouth’ size reflects other dimensions of the feeding
apparatus, larvae may simply be eating all particles small enough to be
swallowed; smaller larvae having a more limited size capacity than larger ones.
The role of food particle size per se can relatively easily be tested by experiment. I
have done this using artificial ‘latex’ spheres which, being of the same material,
eliminate factors other than particle size from the experiment.
Data in Fig. 3 show the range of sizes of ‘latex’ particles fed to and taken by
three dipteran species. The range offered has been chosen to include maximum
sizes eaten by each species in the field. Chironomus eats virtually everything
A. J. McLACHLAN
214
Environment
Chiranomus
Pof$pedi%im
Dosyheleo
I
i
I
I
1
I
25
50
75
I
too
Diameter of 'Latex' spheres ( p m )
Figui.(. 3 . 7 . 1 1 ~tliaiiictcrs ol'artitirial spheres offered to midge larvae in the laborator): and the range
oi'sphere sizes selected by the larvae. Arithmetic 95%C.L. from 60 sample units (horizontal bars) and
C
behveen vertical bars) are given.
IIIC pdi.li( siir I - A I I ~ ldistance
available, the range of particles being remarkably close to that supplied. Both
Polypedilum and Dasyhelea however, take a restricted range. The particles eaten are
t h e smaller ones and are characteristic of the species, Polypedilum commonly
taking microspheres up to about 49 pm in diameter and Dasyhelea those up to
about 42 p m in size.
These results are based entirely on a study of final instar larvae. There is every
reason to believe that the principles concerning mouth size and food item size
will apply to all instars. Whatever the larval age, it is likely, therefore, that larger
lood items are avoided because they are too big to be conveniently swallowed.
However, the possibility that other factors, less easily verified by experiment, may
also be important in determining diet cannot be excluded. For example, field
observations show a main type of particle not eaten, that is the green algae, to
occur as a film apparently confined to the water surface. Even dead algal cells
seem t o decompose in situ rather than sinking onto the mud. These cells may
chct~elbrc.occur in a part of the pool normally inaccessible to the mud dwelling
dipteran larvae.
DISCUSSION
Three types of rock pool occur in predictable places and have typical
tliinciisions. Depth and surface area are especially relevant dimensions as they
cicwrniine the duration of the pool's existence after each rain storm. In contrast
t o data on these dimensions, the available information on pool 'life' is
li-agirtentary. My observations suggest that shallow pools with a large surface
area last about 24 1 1 . A small surface area increases the life to about two days.
D ( ~ p rpools
r
last about a week. Whatever the type of clirnaLe it would seem that
i i o t so iitucli size but life-span of the pool is chiefly responsible for its ecological
(4 i a i x c , c c . r i s t ics.
The predominance of a single species of midge in each kind of pool,
and at extraordinarily high densities, has also been observed by Dr V. Smith
working in Northern Nigeria (pers. commnf and appears to be a feature of rock
TROPICAL RAIN POOLS
215
pools throughout Africa. Densities in the case of one of the animals, larvae
01' t he inidge Po(ypedi1um uanderplanki, reach two million individuals m-*, which
exceeds the highest value of 100 000 larvae m-2 found in the literature. The latter
value was obtained by Konstantinov ( 1958) under special culture conditions for
piduction of midge larvae (Chironomus plumosus) for fish food. With the high
densities and water temperatures up to 41OC a study of secondary production of
rock pools might produce some surprises.
As noted elsewhere in the world (Macan, 1961; Hartland-Rowe, 1972;
Williams, 1975) predators like leeches, dragonfly nymphs and dytiscid larvae are
sciu'ce i i i tcinporary waters. It would, of course, require a predator specialized to
such unstable habitats (reviewed by Southwood, 1977). Because of the high
c1cgrc.c of' mobility normally required by such predators they are easily
ovcrlooketi but include, in tropical Africa, the freshwater crab Potamonautes and
wiiigccl bugs such as the water scorpion Nepa. These 'travelling' predators
iiiigrate ti.oin pool to pool as the water dries up and, presumably, also as the prey
itrc tlcciiiiated. Ranatra and Notonecta are temperate counterparts (Belke 8c Cole,
1975). A study of the diet and impact of these large predators would be
iwvartiiiig.
Tlic location of the pool determines its characteristic type of organic matter.
l'tiis is clearly so in the case of Chironomus imicola pools which are ideally situated
to receive h i t s and blossoms from surrounding trees. Since the filling of these
pools coincides with the annual flowering season, larvae have virtually unlimited
;~c(.css t o this forin of organic matter. The situation is possibly correctly
coiiip;ircd with that experienced by detritivores inhabiting pools and streams in
equatorial rain forests. Here fruits and flowers from surrounding trees are
available and are presumably also eaten throughout the year (Fittkau, 197 1 ) . In
coiitrast detritivores in many temperate waters must depend largely upon the
sporadic autuninal leaf fall for food (Minshall, 1967; Likens & Bormann, 1974;
McLachlan, 19 7 8).
Botli Polypedilum uanderplanki and Dasyhelea thompsoni pools, on the other hand,
:IIT located on the top of bare 'whale-backs', often several hundred metres above
ttic ti'ees. I t would be anticipated that allochthonous organic matter would be
sc;trce, a d this is the case in the P. uanderplanki pools. What is unexpected is the
presence of allochthonous matter in the form of seeds, rodent hair and bones in
D. thompsoni pools. I have seen photographs and footprint casts obtained by Dr M.
A. Cantrell which show that the African civet, Ciuettictis civetta Cabrera and the
genet, Genettafelina Thunberg are responsible. Feeding on rodents and seed crops
at night they return to caves on the whale-backs in the morning and deliberately
dcliute in D. thornpsoni pools, presumably because of their convenient location
i i l l d diiiiensions. The faeces contain the allochthonous material so characteristic of
the sedimcnts i n D . thompsoni pools. Roberts (1954) records the use of selected
delecation spots by these two animals and other members of the family Viverridae
such a s the yellow mongoose, Cynictis penicillata G. Cuvier. It is probably a
wiclespread practice in the family, Winds have been shown to play a similar role in
transporting organic detritus, as well as insects and pollen, to arctic and high
altitude lakes remote from sources of terrestrial production (Mani, 1962). Dung
deposited by elephants on the shores of lakes and reservoirs such as Lake Kariba
(McLidilan, 1974), the droppings of tortoises and seagulls in temporary waters on
Aldabra island (Donaldson & Whitton, 1977) and seals in the temporary pools of
216
A. J . McLACHLAN
Alitarctica ( D r H . J . G. Dartnall, pers. cornm.) are other recently recorded
cxallrples.
7'0 what extent the characteristic types of organic matter associated with each
of' the t h e e kinds of pool influence'the species of dipteran larvae present is of
iiircrest. I t is possible that the massive input of viverrid faeces into one type of
pool tvould exclude a less robust species than D . thompsoni. However, other
lac-tors, notablv pool duration, will probably prove to be more important. In any
cvent ttrc superabundance of organic rnatter'makes i t unlikely that food would be
a liictor liiniting the number of larvae present. Perhaps surprisingly, plentiful
ti~od
seems to be a feature of temporar): rain pools in general, including those in
teinperate regions (reviewed by Hartland-Rowe, 1972).
1 1 1 genel-a1 the food present in the guts of larvae simply represents the most
c.asilv accessible coininon particle. There is no reason to suppose that these items
arc. r&iuireti elements in the diet o r that, other things being equal, larvae could not
d o cyuallv well o n common foods present in rock pools that they d o not normally
i 11h ;I t > i t .
Rock pools are quite difyerent from the better known temporary waters on soft
substrata (reviewed bv Beadle, 1974). Although direct comparison is difficult
lxcause of different inethods of expressing results, my field observations and
data gi\.en bv Rzoska ( 1961) suggest that the densities of animals in pools on mud
;LIT gcneralk verv rnuch lower. In addition, insect larvae iespeciaily dipteran
laiyae! prctf&ninate in rock pools whereas Crustacea (typically euphyllopods)
appear to be typical of pools on soil.
01'the Afiican rock pool Diptera, Dusyhelea is a ubiquitous genus occurring in
siinifar ponds in both Australia and Europe (Disney, 1974; Bishop, 1974).
(,'tiironornus is a genus of opportunist species taking advantage of a variety of
teniporary and otherwise inhospitable freshwater habitats. Polypedilum on the
other hand is not usually a temporary water inhabitant. Polypedilum uanderplanki
appears to be unique in this, as is its ability to tolerate almost complete loss of
t)odv water as a larva, rather than in the egg (Hinton, 1968). I t occurs only in
~ - o c . kpools in tropical Africa.
ACKNOWLEDGEMENTS
Thanks arc due to the University of Malawi and to Professor M. Parr in whose
tfepart~xient most of this work was carried out during an 'Inter-University
Council' Visiting Lectureship in 1976. The Zoology Department, University of
Reading and the Biology Department, University of Ife, made the Nigerian studies
possible, Dr B. Whitton undertook the identification of Ceratopogonidae, Dr V.
G. F. Smith provided a useful correspondence on rock pool ecology and Dr M. A.
C;iiitrell g a \ assistance
~
with the organizational aspects of the programme. Ms F.
Rogers and MI- M . Bafuta assisted me in the laboratory. Mr B. H. Nielsen
(supported by the Danish International development agency) undertook the
cxpcriniental work concerned with particle size selection by rock pool dwellers.
Pi.olessor L. C. Beadle and Dr M. A. Cantrell criticized the paper in manuscript. I
am especially indebted to M i S . M. McLachlan for assistance in the field and for
help with preparation of the manuscript.
TROPICAL RAIN POOLS
27 7
REFERENCES
BAILEY, M. T. J . , 1959. StatisticalMethodsinBiology. London: English Universities Press.
BEADLE, L. C., 1974. The inland Waters ofTropica1 AJrica. An Introduction to Tropical Limnology. London: Longman.
BELKE, D. & COLE, G; A . , 1975. Adaptational biology ofdesert temporarypond inhabitants. In M. L. Hadley
(Ed.), Environmental Physiology $Desert Organisms: 207-226. Philadelphia: Dowden, Hutchinson & Ross,
BISHOP, J. A., 1974. Tfiehunaofternporaryrainpoolsin Eastern NewSouth Wales. ffydrobiologia, 44.319-323,
CIJMMINS, K. W. & WUYCHECK, L. C., 197 1. Calorific equivalents for investigations in ecological energetics.
Verhandlungen der Internationalen Vereinigungfiir Theoretische und Angewandte Limnologie, Mitteilungen, 18: 1-1 58.
CIJMMINS, K. W., MILLER, L. D., SMITH, N. A. & FOX, R. H., 1965. Experimental Entomology. London:
Rhiiiclioltl.
DISNEY, R. H . L., 1974. A midge (Dipt: Ceratopagonidae) new to Britain that is abundant in the limestone
p.iveniciit 01 the Yorkshire Pennines. Entomologists'Monthb Magazine, 110: 227-228.
DONALDSON, A. & WHITTON, B. A., 1977. Chemistry ofFreshwatet. Pools on Aldabra. Atoll Research Bulletin No.
2 13. W;isliingtoii, D.C.: Smithsonian Institute.
DOWGIALLO, A., 1975. Chemical composition of an animal's body and its food. In W. Crodzinski, R. 2.
Klenowski &A. Duncan (Eds), Methodsfor Ecological Bzo-energetics: 160-199. IBP Handbook No. 24. London:
B1;ickwcIl.
b ITTKAIJ, F. J., 197 I. DiatributionandecologyofAmazonianchironomids. Canadian Entomology, 103: 407-413.
t IARTLAND-ROWE, R., 1972. The liinnology of temporary waters and the ecology of Euphyllopoda. In R. B.
Clark'& R. J . Wooton (Eds), Essays in Hydrobiology: 15-31. Exeter: James Townsend & Sons.
H INTON. H. E., 1968. Reversible suspension ofmetabolism and the origin oflife. Proceedings oJtheRoynlSociet~~if
London, ( B ) 171: 43-57.
HYNES, H . 6. N. & KAUSHIK, N. K., 1969. The relationship between dissolved nutrient salts and protein
production in submerged autumnal leaves. Verhandlungen der Internationalen Vereinigungf u r Theoretische und
Angewandte Limnologie, 17: 95-103.
JONES, R. E., 1975. Dehydration in a n Australian rock pool chironomid larva (Parabomiella lonnoiri).Journal of
~!ll/fJt/UJ/i~
( A~ ~) ' i, y : 1 1 1-1 19.
JUNK, J. W., 1970. Investigations on the ecology and production biology of the 'Floating Meadows' (Pa.Tpa/ol : c . h i / i i j t h l o v l i L t r r ) on the iniddle Amazon. hazoniana, 2: 449-495.
KONSTANTINOV, A. S . , 1968. Cultivation ofchironomid larvae. Translated from the Russian. Fisheries Resrarrh
/lomiij/,/Cannda, TrarislationmieJ No. 2133:5 7 pp.
LIKENS, G . E. & BORMANN, F. H., 1974. Linkages between terrestrial and aquatic ecosystems. Bioscience, 24:
,447-456.
LOWRY, 0 . H., ROSEBPUCH, N. J., FARR, A. L. & RANDELL, R. J., 1951. Protein measurement with the
Folin phenol reagent.JournalofBiologica1 Chemistry, 193: 265-275.
MACAN, T.T., 1961. Facl'tors that limit therangeoffreshwateranimals.BiologicalReviews,36:151-198.
MANI, M. S., 1962. lnlroductionto Nigh Alkitu.de Entomology. London: Methuen.
MtLACHLAN, A. J., 1974. Development ofsoine lakes ecosystems in tropical Africa, with sprcial reference to the
iiivc.rtcl)i.,ites.Biological Reviews, 49: 365-397.
M (LACHLAN, A. J., 1978. Interactions between freshwater animals and microorganisms. Annals of Applied
/ ~ l i l / i J g Y ~89:
,
162-165.
hl(.l.ACHLAN,A.J., BRENNAN,A.& WO'ITON, R. S., 1978. Particlesizeandchironomid(Diptera)foodinan
upland rivcr. O i k o ~3, 1 : 247-252.
M INSHALL, G. W., 1967. Role of allochthonous detritus in the trophic structure of a woodland spring brook
conimnnity. I:'culogy, 48: 139-149.
MOOR, G. W., 1970. Lininological studies of temporary ponds in South-eastern Louisiana. The Southweslrrn
,Va/iira/i.d. 15: 83-1 10.
REISS, F., 1976. Die Benthoszoozonosen zentralamazonischer Varzeaseen und ihre Anpassungen a n die
jahresperiodischen Wasserstandsschwankungen. Biogeographica, 7: 125-135.
ROBERTS, A,, 1954. The Mammal4 of South AJrica. Cape Town: Trustees of'The Mammals ofSouth Africa' Book
Fuiid.
RZOSKA, J., 196 I . Observations on tropical rain-pools and general notes o n temporarywaters. Hydrobiologia, 17:
268-28 1.
SOllTHWOOD, T. R. E., 1977. Habitat: the templet for ecological strategies?Journa/ ofAnima/ Ecdogy, 346:
3 s 7-365.
WILLIAMS, W. D., 1972. The uniqueness of salt lake ecosystems. In Z. Kajak & A. Hillricht-Ilkowska (Eds),
P ~ o d i ~ i / i 7Problem
~it~
in Fre~hwaters:349-361. Proceedings of the IBP-UNESCO Symposium 1970. Warsawa:
PWN Polish Science.
WILLIAMS, W. D., 1975. A note on the macrofauna o f a temporary rainpool in semi-arid Western Australia.
A ic\lrcilian,/ourna/of Marine and Freshwater Research, 26: 425-429.