Odonalologica33(2): 129-146 June I. 2004
Adult Zygopteraof Kibale NationalPark, Uganda:
habitat associations and seasonaloccurrence
* ’² C.K. Apodaca¹ andL.J. Chapman¹
'Department ofZoology, University of Florida, Gainesville, FL 32611, United States
2 WildlifeConservation Society, 185th Street and Southern Blvd., Bronx, NY, 10460,United States
Received February22, 2003 / Revised and AcceptedNovember 3, 2003
In this study, a 10-month survey of four aquatic sites in Kibale National Park, Uganda
was used to quantify seasonal and spatial variation in both limnologicalfeatures ofthe sites
and adult damselfly assemblage structure. Of the 4 limnological characters measured dis-
solved the variable from of 1.01 I oxygen was most among sites, ranging an average mg 1
in the interior of the Rwembaita 6.71 I’ in Swamp (a papyrus-dominatedwetland) to mg 1
ofthe richness similar sites and did an inflowingtributary swamp. Species was among not
correlate with dissolved oxygen concentration. However, site was a significant predictor of
occurrencefor some spp. This suggests that site effects are important, and that a combina-
tion ofsite-specific environmental characters may underlie the observed distributional pat-
terns. Seasonal fluctuation in rainfall was not a goodpredictorofZygoptera activity. Several
active in both the and adult Proischnura spp. were wet dry seasons. Surprisingly, subfur-
detected in the ofthe Rwembaita catum were year-round hypoxic waters (papyrus) Swamp
and did other sites in the larval adult that this is notoccur at any or phase, suggesting sp. a
swamp specialist.
INTRODUCTION
A central theme of ecology is understanding patterns in the distribution and abun-
danceoforganisms and the. selectivepressures that underliethese patterns. Clearly, the
abiotic environmenthas a major influenceon theecology of organisms (WOLANSKI
& GERETA, 2001). Abiotic factorsmay affect growth,reproduction, and survivorship,
thereby influencing patterns ofdistributionand habitatuse (FLEMER et al., 1999). For
aquatic insects, important abiotic conditions include water temperature (ARTHUR et
al., 1982;GILLOOLY & DODOSON, 2000), dissolved oxygen concentration (DAVIS,
* Corresponding Author: Fairbanks Fish and Wildlife Field Office, 101 12th Ave., Box 17, Room 222,
Fairbanks, AK 99701-6236, United States,phone; 907-456-1874; e-mail: [email protected] 130 C.K. Apodaca & L.J. Chapman
1975; JACOB et al., 1984), pH (HALL& IDE, 1987; ROUSCH et al„ 1997; COURT-
NEY, 1998), flow (FEMINELLA, 1996; WELLNITZ, 2001), and conductivity (CAN-
NINGS & CANNINGS, 1987) since it is reflective oftotal dissolved ionconcentrations
ofthese factors habitats (ALLAN, 1995). Many vary amongaquatic (i.e., streams, lakes,
and fluctuate (HUBERTZ & CAHOON, 1999, FENCHEL, swamps, etc.) may daily
2000) or seasonally (LAHR et al., 1999; BORSUK et al, 2001). The role these factors play in shaping habitat quality may explain patterns of geographic distribution, habitat associations, and seasonal variation of some aquatic insects.
Zygoptera have a two-stage lifehistory pattern, and characteristicsof both aquatic and terrestrial environments shape the ecology of this group. With few exceptions (WIL-
LIAMS, 1936; KELTS, 1979; WINSTANLEY, 1983;GARRISON &MUZON, 1995) damselfliesinhabitfresh in andlarval their waters egg stages priorto metamorphosis into
terrestrial affects a winged adult. Thus, the aquatic environment egg and larval stages directly and affects adults indirectly due to selective pressures to choose favorable sites for egg and larval development. For example, odonateeggs and larvae may suffer high mortality, or slow or impaired development when exposed to unfavorable conditions such as desiccation(BENNETT & MILL, 1995) or sub-optimal temperatures (LAW-
TON, 1970; HALVERSON, 1983; LUTZ & ROGERS, 1991; HAWKING & NEW,
these conditions adult and 1995). Because may vary seasonally, activity breeding may be timed to coincide with optimal environmental conditions in the larval habitat.
Patterns of adult activity, breeding, and larval development in odonate species dif- fer between and In larval markedly temperate tropical regions. temperate regions, pe- riods (time between hatching and emergence) tend to be longer than in tropical species
(CORBET, 1980; KUMAR, 1976). Mosttemperate damselfliesrequire 1 to 2 years for a complete generation (CORBET, 1999), and some may take more than 4-5 years to complete larval development (NORLING, 1975; UBUKATA, 1980). In contrast, some tropical damselflies havethree or more generations per year (KUMAR, 1976). Because growth indamselflies is temperature-dependent (PROCTER, 1973), differencesin larval
between and reflect latitudinal vari- periods temperate tropical regions are thought to ation in climatic conditions, specifically temperature (CORBET, 1999). Furthermore, broad-scale zoogeographical patterns in species distributions also reflect isoclines of temperature variation. In temperate regions, seasonal fluctuations in photoperiod are also very important to the ecology of larval dragonflies and damselflies, particularly
for in as cues timing larval development (CORBET, 1955; PROCTER, 1973). In the tropics, where seasonal variations in temperature and photoperiod are more modest,
rainfall patterns of seem to be more important in controlling seasonal patterns of larval
affects development (KUMAR, 1972), though temperature ultimately developmental rate (KUMAR, 1976).
Whiletropical systems are known to support impressively diverse insect faunas (ER-
WIN, 1982), the majority ofthese groups are poorly understood,and studies linking sea-
sonalvariationin abiotic conditionsto patterns ofzygopterandistributionand abundance
limitedin and few studies are number geographic scope. In particular, there are Unking Zygoptera of Kibale National Park, Uganda 131
distributionoflarval damselflies with characteristicsof the aquatic environmentin Af- rica. this attributed lack of In someregions may be to the larval taxonomy. For example,
in East few larvae have been linked adult forms. This taxonomic Africa, zygopteran to
constraintlimits ourability to develop predictive modelsof damselfiy larvaldistribution based on aquatic habitatcharacters. However, habitatrelationships may be cautiously
inferred based on distributionand behavior ofadults, because seasonal patterns of adult
activity often reflect patterns in the timing of larval emergence, adult longevity, and
emigration (CORBET, 1999). In this study, we use this tool to understanddistribution
in patterns and assemblage structure an Afrotropical damselfiy community.
In general, the ecology ofEast African damselflieshas not been extensively described, and patterns of habitatassociation and seasonal variation remainlargely unknown. How-
recent distributional in East Africa have ever, surveys (CLAUSNITZER, 2001) pro-
vided preliminary data for diversity hot spots in the region (e.g., wetlands and swamp
forests). Forty-seven odonate species are found in Kibale National Park, Uganda and
the surrounding areas (D1JKSTRA& DINGEMANSE, 2000). These preliminary sur-
veys suggested that damselfiy assemblage structure varies among habitats. These data
in with data collected in 1999 raised the conjunction preliminary (C.K. Apodaca) pos-
sibility that the papyrus swamps of Kibale National Park may host a unique damselfiy
These abiotic conditions because assemblage. swamps provide challenging they are
characterized by chronically low oxygen throughout most ofthe year (CHAPMANet
al., 2000). The objective of this study was to describe damselfiy species assemblages
across four aquatic sites inKibale NationalPark, Uganda that differin basic limnologi-
cal features to detect links between adult assemblage structure and the habitat charac-
teristics of the aquatic environment.
In this paper, we present results from a 10-monthsurvey ofadult damselflies at four
sites inKibaleNational Park. with the measured Concurrently surveys, we conductivity,
dissolved oxygen, pH, rainfall, and water temperature at these sites. Our specific objec-
tives were to (1) describe species assemblages at fouraquatic sites (two swamp andtwo
stream sites),(2) describe seasonal variationin adultdamselfiy occurrence at thesesites,
and (3) provide preliminary information on seasonal variation in breeding behavior of
adults. We predicted (1) the use of swamp habitats by damselflies follows a seasonal
with in the (2) fit into pattern, most activity occurring wet season, many species com-
that are to certainhabitats, and the of adult munity assemblages specific (3) presence damselflies correlates with limnological characteristics of the larval environment.
STUDY AREA AND METHODS
o STUDY AREA. - Kibale National Park (0°13’-0 41’N and 30° 19’-30°32’E)is located in western Uganda
nearthe foothills ofthe Rwenzori Mountains. Approximately 58% ofKibaleis characterized by semi-decidu-
and forest & LAMBERT, with from 25 30 ous evergreen (CHAPMAN 2000), canopy heightranging to m
(BUTYNSKI, 1990). The remainder ofthe park is secondary forest (19%),grassland (14.5%), wetlands (2%),
woodland (6%), and plantations (1%; CHAPMAN & LAMBERT, 2000), Kibale NationalPark receives ap-
of proximately 1749 mm rainfall annually (1990-2001) that peaks duringtwo rainy seasons(L.J. Chapman 132 C.K. Apodaca & L.J. Chapman
& C.A.Chapman, unpublisheddata)with most rain fallingMarch through May and Septemberthrough No- vember. the varies The However, the onset, duration,and magnitudeof rains among years. mean daily mini-
is 15.5 and the maximum is mum temperature °C; mean daily temperature 23.8 °C (1990-2001).
SITE DESCRIPTIONS AND LIMNOLOGICAL SAMPLING. - We quantifiedseasonal occurrence of adult damselflies and characterized seasonal fluctuations in four limnologicalparameters at two intermittent forest streams Stream (Mikana andInlet Stream East), a largepapyrus swamp (the Rwembaita Swamp), and a permanently flowing swamp stream (Njuguta Swamp Stream)in Kibale National Park (Fig. 1).
Four characters dissolved and limnological (conductivity, oxygen, pH, water temperature) were sampled
the four sites where the ofadult damselflies monitored. bi-monthly at seasonal distribution was Dissolved
and measured with Yellow oxygen (mg I 1 ) water temperature (°C) were a Springs Instruments (YSI) meter
(Model95). Conductivity (pS cm 1 ) was measured with a YSI Model 30 meter, and a chek-mite pH sensor
used to Rainfall data recorded the Makerere Field was measurepH. were at University Biological Station
(Fig. 2), located approximately 3 km from the Rwembaita Swamp. Water current was ranked on a scale from
0 to 3 (0 = nocurrent, 1 = visible but low flow, 2 = fastflow with little water perturbation,3 = fast flow with water perturbation).Detailed descriptionsofeach study site are provided below.
INLET STREAM — EAST. This stream flows throughswamp forest and then throughmature secondary forest before the The site 220-m stretch ofthe before its confluence entering swamp. study was a stream just
ISE 25 in with the papyrus swamp. During the period of study averaged cm depth and 1 (on a scale of0 to
in flow. Dissolved and measured four 3) oxygen water temperature were at stations at this site, and conduc- tivity and pH were measured from water samples taken at three ofthese stations. Collection of damselflies occurred over the 220-m study stretch.
MIKANASTREAM. — This intermittent and stream is 1,400 m long passes through mature secondary forest before flowing into the Rwembaita This is the of the Swamp. stream largest tributary swamp, aver-
Fig. 1. Map of the study areain Kibale National Park, Uganda, East Africa. Site 1: Mikana Stream; — Site — 2: Inlet Stream East; Site 3: Rwembaita Swamp; — Site 4: Njuguta SwampStream. Zygoptera of Kibale National Park, Uganda 133
aging approximately 1 m in width
(CHAPMAN et al„ 2000). Dis- solved and oxygen water tempera- ture were measured at 13 sampling stations distributed along a 275 m reach starting at the Census Road bridgeand extendingdownstream.
During the study periodthese sam- pling stations averaged 17 cm in depth and 0.8 in flow (on scale of
0 and to 3). Conductivity pH were measured from three water samples taken at the bridge and 205 m and
255 m downstream. For each sam- pling periodaverage water temper- Fig. 2. Monthly rainfall (mm) measured at the Makerere Biological ature (n= 13),dissolved oxygen (n Field Station, Uganda throughout the study period (August 2000 to = 13), pH (n= 3),and conductivity June 2001). (n = 3) were used to characterize the site. Collection of adult damselflies occurred from the bridge to 275 m downstream.
NJUGUTA SWAMP STREAM. — In the northern region of Kibale National Park, the hypoxic water from the Rwembaita Swamp flows into the well-oxygenated Njuguta River by way of this short connect-
3 in reaches ofthis shaded ing stream (approximately m average width). Upper stream are partially by over-
and All hangingtrees vegetation. limnologicalcharacters were measured at four stations spaced 20 m apart
ofthe River For each between 20 to 80 m upstream Njuguta confluence. samplingperiod, the average water
dissolved and of the four stations used characterize the site. temperature, oxygen, pH, conductivity were to
During the study period these samplingstations averaged 129 cm in depth and 1.3 (ona scale between 0 and
in flow. Adult from the where the flows the 3) damselflies were collected point swamp stream into Njuguta
River up to 100 m upstream ofthe confluence.
— is fed several forest but RWEMBAITA SWAMP. This large papyrus-dominated swamp by streams,
with few channels of during drier months of the year the swamp consists mostly oflarge stagnant pools a
running water. During the wet season, the influx ofrunning ground water transforms the interior swamp to
network of channels. characters measured in ofthe interior six a flowing Limnological were pools swamp at
stations. For each the dissolved and sampling period, average water temperature, oxygen, pH, conductivity
the stations ofthe six stations were used to characterize the site. During study period, these sampling aver-
aged 55 cm in depth and 0.3 (on a scale of 0 to 3) in flow. Adult damselflies were collected near stagnant
and channels The collection extended about pools along offlowing swamp water. area 100 m into the inte-
rior swamp from the swamp bridge.
COLLECTION AND PRESERVATION OFADULT DAMSELFLIES. - Adult damselflies werecol-
lected bi-monthly at the four study sites within Kibale National Park between August 2000 and July 2001.
Two people with aerial nets visited each site and actively searched for adult damselflies for approximately
1 h. While collecting, the observers actively searched for females and males of all species present. Captured
damselflies were immediatelypreserved in 95% ethanol alcohol and subsequently identified with the aid of
dichotomous We tried collect adults the a key (V, Clausnitzer, unpublished). to consistently during warm-
est time in the day. For each sampling period, we recorded presence or absence of males, females, and pairs
in tandem for each species.
- in either the STATISTICAL ANALYSES. For all analyses, samples were categorized as occurring
rainfall wet or dry season.The dry season(Novemberto March,and June) was defined as months when was
below the median, and the wet season (August to October and April and May) included all months when
above the median. of and rainfall was Analysis ofvariance (ANOVA) was used to compare means wet dry
for dissolved and the When seasonmeasures conductivity, oxygen, pH, water temperature among four sites.
means differed among sites, we used Scheffe’s post-hoc test to perform multiple comparisons. Pearson cor- 134 C.K. Apodaca & L.J. Chapman
relation was used to detect relationshipsbetween limnologicalcharacters over the study period within sites
rank =22 correlation used detect characters (n per site). Spearman was to relationships among limnological across the four sites.
To the of adult sites and between used bi- compare probability damselfly occurrence among seasons, we
the nary logistic regression. For each species, we modeled probability ofoccurrence duringa sampling pe- riod and as a function ofsite (nominal,four sites), season(nominal,wet dry), and the interaction term. Each
treated in We with all factors in themodel and then survey was as an independentsample our analysis. began used the Wald statistic to perform backward stepwise elimination until only significant variables remained in the model (NORLlSlS/SPSS, 1999).Logistic regression uses maximum likelihood algorithms to estimate the of when variables in- probability occurrence.However, sample sizes are small or categoricalpredictor
clude oneor more levels with no observations, maximum likelihood estimates are unreliable. This occurred when damselflies were uncommon (< 4 total detections) orwere entirely absent from some sites (e.g., Pro- ischnura subfurcatum). For three species (Ceriagrion glabrum, Chlorocnemis marshalli superba, Africal- lagma pseudelongatum), there were fewer than four detections throughoutthe study. Data for these species were not included in the analysis due to the small sample size. In these cases, we describe patterns of distri- bution qualitatively.For four other species tenuis C. trifaria, Proischnura subfurcatum, and (Chlorocypha ,
because these occurred sites Pseudagrionspernatum ) logistic regression was problematic, species at some but absent other To model observation were entirely from sites. facilitate convergence we added a single to the sites in which those detected.This species were not simple adjustment allowed usto estimate differences
and did estimates ofseasonal effects.All statistical were conducted SPSS among sites, not change tests using
and results considered P 0.05. (10.0), were statistically significant at <
LIMNOLOGICAL CHARACTERS
During the survey period, bi-monthly rainfall ranged between 1 mm and 253 mm.
The mean monthly rainfall during the wet season (mean = 256) was greater than during
= differed the dry season (mean 98). Limnological parameters also between seasons.
In general, conductivity was higher during the dry season and lower in the wet season
(Fig. 3). Dissolved oxygen also variedbetween seasons, but the magnitude ofthis vari- ation differed among sites (Fig. 3). The most extreme variation in dissolved oxygen oc- curred attheNjuguta Swamp Stream, where wet season levelsneared saturationand dry
season levels dropped to 1.94 mg I 1 (Fig. 3). Water temperature and conductivity var-
ied little between seasons. Within sites, there were few significant correlations among
correlated with limnological characters. However, conductivity was negatively water
Rwembaita = -0.57, = 0.010) and Inlet Stream East temperature at the Swamp (r p (r
= -0.64, p = 0.003).
Some limnological characteristics also differed amongthe four study sites. Dissolved oxygen and pH differedamong sites in both the wet (dissolved oxygen: P < 0.001;
pH: P = 0.006) and dry seasons (dissolved oxygen; P < 0.001; pH: P < 0.001; Tab. I).
Dissolved oxygen and pH were lowest in the Rwembaita Swamp and highest in Mi- kana Stream in both seasons (Scheffe, P < 0.05, Tab. I). Mikana Stream did not differ
from Inlet Stream East or the Njuguta Swamp Stream in wet season dissolved oxygen
levels. Conductivity and water temperature did not differsignificantly among sites in either season (Tab. I). Across four sites, pH was positively correlated with dissolved
= 1.0, P < 0.01 oxygen (r ). s Zygoptera ofKibale National Park, Uganda 135
SPECIES DISTRIBUTION
During the study period, we observed 10 species of adult damselflies belonging to four families (Tabs IIa,b). The damselfly assemblages were very similar with respect to richness and seasonal occurrence at Inlet Stream East and Njuguta Swamp Stream.
Very few species were unique to any one site. The exceptions were Chlorocypha trifaria,
which was only recorded at MikanaStream, andProischnura subfurcatum, which is a
common inhabitantofthe RwembaitaSwamp throughout the year. Below is a synopsis
of distributioninformationfor each species observed during the study.
Fig. 3.Mean valuesof four limnologicalcharacters [conductivity (pS cm 1 ), pH, water temperature (°C), and dis-
1 solved oxygen concentration (mgl )] measured throughoutthe study period (August 2000 to June 2001). 136 C.K. Apodaca & L.J. Chapman
Table I
Mean values for characters for four sites in in limnological a swamp-river system Kibale National Park,
sites detected with Uganda. Differences among were using ANOVA Scheffe contrasts (underlined sites are
Sites coded follows: R Rwembaita not significantly different). are as = Swamp, N = Njuguta Swamp Stream,
= M I Inlet Stream East, and = Mikana.
Mean Scheffe
Character Season F P contrasts
PH Dry 6.67 63.58 <0.001 RNJM
Wet 6.61 4.94 0.006 RNIM
Conductivity Dry 53.64 3.97 0.015 NRMI
(jjS cm') Wet 54,69 1,79 0.168 NR1M
76.13 <0.001 Dissolved oxygen Dry 4,08 RNJM
(mg 1') Wet 4.58 52.21 <0.001 RINM
Water temperature Dry 16.93 3.79 0.18 MRIN rc) Wet 17,54 0.401 0.754 MRNI
COENAGRIONIDAE
AFRICALLAGMA PSEUDELONGATUM (LONGFIELD). - There were not enough detections
of to statistically analyze predictors occurrence. Maleswere only captured duringsurveys, but observed in tandem in 2001 times. mating pairs were July during non-survey
CERIAGRION glabrum (BURMEISTER). - Altogether we recorded three individuals during our study. In August of2000 (n = 1) and 2001 (n = 2) pairs in tandemwere seen in the interiorof RwembaitaSwamp. In August 2001, a malewas captured while it was eating an adult male Proischnura subfurcatum. There were not enough detections of this species to statistically analyze predictors of occurrence.
PROISCHNURA SUBFURCATUM (SELYS). - This was the only swamp specialist found within Kibale NationalPark. Logistic regression indicated site as a significant predictor for this species (Tab. III). This can be attributed to theabsence ofthis species at all sites other than the swamp. Males were detected throughout the study period, and females were found during some ofthe dry (October, November, and June) and wet (August,
October, April, and May) months. Copulation was observed only during the dry season
(November, March, and June).
PSEUDAGRION HAGENI TROPICANUMPINHEY. - Logistic regression indicated that the
ofits varied sites Examination probability occurrence among (Tab. III). of parameter estimates indicated that when compared to Rwembaita Swamp the probability of oc- currence was greater at Inlet Stream East and Njuguta Swamp Stream and less at Mi- kana Stream. In the Rwembaita Swamp this species was most common near channels
ob- offlowing water in the interior swamp. At the Njuguta Swamp Stream males were
served mate-guarding females during oviposition. Males continued to clasp females and completely submerged themselves below the surface of the water with the female during oviposition. Zygoptera ofKibale National Park, Uganda 137
PSEUDAGRION KIBALENSELONGFIELD. — Logistic regression indicatedthat the prob-
ability of its occurrence at Inlet Stream East and the Njuguta Swamp Stream was the
same as Rwembaita Swamp, but greater at Mikana Stream (Tab. III). It was caught in
tandem atMikanaStream during both dry (December, January, and June)andwet (April)
months, and also at RwembaitaSwamp in the wet season (April).
PSEUDAGRION SPERNATUM SELYS. — Logistic regression indicatedno difference in
the of this the sites it its probability detecting species among occupied. However, occur- rence differed between seasons (Tab. III). Species detectionswere less probable during
the wet season and greater during the dry season. It was captured in tandemonly during
the driermonthsof the year at the Njuguta Swamp Stream (December, February, and
March) and in the RwembaitaSwamp (August and June).
CHLOROCYPHIDAE
CHLOROCYPHA TENUIS LONGFIELD. — Logistic regression indicate site as a signifi-
cant predictor for the presence of this species (Tab. HI). This may be attributedto the absence of this species at the Rwembaita Swamp. Examinationof parameterestimates
indicated that when compared to the RwembaitaSwamp, the probability of occurrence
was greaterat Inlet Stream Eastand Njuguta Swamp Stream, and less at Mikana Stream.
Most captured individuals were males. A breeding pair was captured in tandem in the
wet season (August) at Inlet Stream East. Otherwise, females were recorded at Inlet
Stream East during wet (April and May) and dry (June) monthsand at Mikana Stream
during a wet (May) month.
CHLOROCYPHA TRIFARIA (KARSCH). — Logistic regression indicated site as a signifi-
cant predictor for this species (Tab. III). Males were observed in both wet and dry sea-
sons, but only one femalewas captured throughout the study period.
PROTONURIDAE
CHLOROCNEMIS MARSHALLISUPERBA SCHMIDT. — It inhabitssmallforest streams with-
inthe park, but is somewhatuncommon. We recorded only oneadult maleatthe Mikana
Stream. that it low densities other forest We expect occurs at at streams. C. pauli (Long-
field), has been observed withinthe park, but was not observed during this study.
CALOPTERYGIDAE
— included in UMMA SAPHIRINA FORSTER. It was not the distribution analysis, al-
though it was observed at both Mikana Stream and Inlet Stream East between May and
of for August 2000 and 2001. Data were not collected this species during other months
of the study period. 138 C.K. Apodaca & L.J. Chapman
2 2 2 2 Jn M M B Jn — — — — — — Dry Dry — tandem; Jn, M B Jn, T M M 2 2 2 2 MM in My MMMMM M M B My M M M
My,My M M My,My M B T=pairs My, BMBBB My, 2 Wet A,Aj B Wet Aa T M B M
9; A, M M A, M
and 2 Mr, M M Mr,Mr M cJ = Mr,Mr, M M M Mr,Mr, M B
2 M M 2 Ff M M Ff M M M 9; F, M M F= F, M MMMM F,F. M MTMMMMMTMMTM 2 Ja, MMMMMMMM M M M T Ja MMMMMM Ja, indicated] Dry DryDry [M=cJ; Ja,Ja, M M M Ja,Ja, M M
2 ; months DD DD — — — — — —
D, D,D. T M the M M
in N, 3 N, T M n M 2
2 2 M M and Nn M M M Nn M
1
, NN. M M N,N M Ha 2 2 periods 0o M 0o M MMMMMTM Table O,o, M M M O,o, M
2 2 sampling Ss — — — — — Ss Wet Wet to S,s, M M M M S,s,
2 2 M M refers Aa M M Aa -
, etc. A, M M T A,A M
, 2
S S,, Aj, superba A,, 2 tropicanum superba N 2 3 tropicanumtropicanum superba pseudelongatum subfurcatumsubfurcatum kibalense spematumspernatum trifaria marshalli pseudelongatum subfurcatumsubfiircatum kibalense spernatumspematum trifaria marshalli - — glabrumglabrum hageni tenuis trifaria glabrum hageni tenuis Africallagma Ceriagrion ProischnuraProischnura Pseudagrion Pseudagrion Pseudagrion Chlorocypha Chlorocypha Chlorocnemis Africallagma Ceriagrion Proischnura Pseudagrion Pseudagrion Pseudagrion Chlorocypha Chlorocypha Chlorocnemis SeasonalpresenceofadultdamselfliesatInletStreamEastandMikanaStreamfromAugust2000toJune2001 Season Month Species Season Month Species East StreamStream Stream Coenagrionidae Chlorocyphidae Protoneuridae Coenagrionidae ChlorocyphidaeChlorocyphidae Protoneuridae Inlet MikanaMikana Zygoptera of Kibale National Park, Uganda 139
DISCUSSION
This study provides some of the first information on habitat associations and sea- sonal variation in occurrence of an Afrotropical damselfly community in East Africa.
In Kibale National Park, limnological characteristics vary among sites and to a lesser degree between seasons. Our data suggest that inter-site variation in limnological fea-
influences distribution. habitat tures damselfly Two species appear to be specialists,
associated with that characterized extreme being exclusively swamps are by aquatic
One found at the and at two sites where the hypoxia. species was swamp oxygen was not chronically low. Four species occupied at least at one of the sites characterized by moderate to high dissolved oxygen, but were not found in the hypoxic swamp. Two
occurred in all habitat of these species were widespread and types surveyed. Many spe- cies most of with few of seasonal were present throughout the year, examples strong patterns of occurrence being detected.
INTERSITE VARIATION
in this Of the four limnological characters considered study, dissolved oxygen con- centration was the most variableamong sites. Although there was a positive correlation between average pH and dissolved oxygen among the four sites, the magnitude of fluc-
tuationfor these characters differed. Over the course of the study average pH ranged between 6.42 and 6.87 among the four sites. This represents a 7% difference among
ion concentration, + sites the sites. If pH is translatedto hydrogen [H ], the range among
7 7 + was 1.4 x Id to 3.8 x Id M H a 270% difference. Dissolved , representing oxygen
showed even variation sites ranging between 1.01 and 6.71 T l greater among mg , rep- resenting a differenceof 664%.
The wide range in dissolved oxygen availability in this swamp-river system may be an important factor for damselfly distribution, though certainly other limnological
features may also contribute to the patterns observed. Zygoptera, and more generally,
sensitive insect tracheal gill breathers, tend to be more to hypoxia than other aquatic
larvae such as atmospheric breathers (e.g., Nepidae, Dytiscidae, and Hydrophilidae as
larvae and adults) and transportable airstore breathers [e.g., Dytiscidae (adult), Gyrini-
dae (adult), Naucoridae (larvae and adults)]. Tracheal-gill-breathing insects are more
sensitive to low oxygen conditions, because they are benthic (MERRITT & CUM-
MINS, 1996) and extract dissolved oxygen from the aqueous environmentby simple
diffusion (WIGGLESWORTH, 1972). The occurrence of six damselfly species in the
Rwembaita Swamp site, some ofwhich occurred throughout the year, suggests a broad
habitat representation in the Zygoptera and far-reaching adaptations to hypoxia in some
larval zygopteran species.
For four ofthe nine species reported in this study, site was a significant predictor of
occurrence. Proischnura subfurcatum and Ceragrion glabrum were the only species ob-
served exclusively in the Rwembaita Swamp. The adult damselfly community assem- 140 C.K. Apodaca & L.J. Chapman
2 2 2 — — — — Jn — — — — — — — — Jn — 9; Dry Dry and Jn, M Jn, M T T T
2 2 2 B MM AA, — — — — — A,Aj M T Wet — Wet MM T t M M M M AA, A, M BMB F, M M T M p.F, M M M 2001 months Ja, — — — — — — Ja,Jaj MMMMTM MMMMMMM M Dry — June Dry , Dry MMM the Ja,Ja M M M Ja,Ja, M M M to in 2 2 TM 2 T 2000 Dd D° — — — — — — — — — and D,D, M M D, M M M 1 D, M 3 MM August n M M periods M N,N, M 2 2 MMM n — T M from N — — — — — — — — NN M N, M M N, — BMTMMM M lib sampling 2 2 MMM 2 Swamp B 0o M 0o M Table to o, M M O,o, B refers O 2 r Ss M M M s Rwembaita MMMMM S — — — — — — — — etc. Wet Wet Wet Wet , S, , s, M S,s M 2 and S 2 2 A, S,, a — — — — — — — — — AA — — — — — — — — — A,, Stream A, M A, B T A,, Swamp - tandem; superba superba Njuguta in N 2 3 tropicanum 2 tropicanum at pairs = pseudelongatum subfurcatum hageni kibalense spernatumspematum tenuis trifaria marshalli pseudelongatum subfurcatum hageni kibalense spernatumspematum tenuis trifaria marshalli damselflies- — T glabrum glabrum adult Season Month Species Africallagma Ceriagrion Proischnura Pseudagrion Pseudagrion Pseudagrion Chlorocypha Chlorocypha Chlorocnemis Season Month Species Africallagma Ceriagrion Proischnura Pseudagrion Pseudagrion Pseudagrion Chlorocypha Chlorocypha Chlorocnemis of Stream Swamp presence Swamp Seasonal Njuguta Family Coenagrionidae Chlorocyphidae Protoneuridae Rwembaita Family Coenagrionidae Chlorocyphidae Protoneuridae Zygoptera ofKibale National Park, Uganda 141 blage withinthe RwembaitaSwamp is segregated between areas of stagnant and flowing water. Proischnura subfurcatum and Africallagmapseudelongatum were generally found near standing pools of stagnant water, while the other species observed in the swamp, Pseudagrion hageni tropicanum, P. kibalense, and P. spernatum,were mainly foundin or near channelsof flowing water (C. Apodaca, pers. observ.). The latter species were also observed at stream sites where mean dissolved oxygen levels were moderately high 1 for Inlet Stream Eastand 4.3 1 for but (5.0 mg 1 mg 1 Njuguta Swamp Stream), they were seasonally rare (P. hageni tropicanum)ior absent (P. spematum) from MikanaStream, a site characterized by fastercurrent and higher dissolved oxygen levels (mean dissolved 1 oxygen concentration = 6.7 mg I )- Both chlorocyphid species, Chlorocypha tenuisand found Chlorocypha trifaria were only at stream sites, and C. trifaria was restricted to Mikana Stream along with the protoneurid Chlorocnemis marshallisuperba. The other members of the damselfly community were recorded at all four sites. In Uganda, and East Africa in general, few zygopteran larvae havebeen linked toadult the distributionand behavior ofthe adults correlatewith forms; however, may patterns of larval distribution.In Kibale National Park, some zygopteran larvae that have been recorded at various aquatic sites tend to supportpatterns ofadult distribution.For exam- ple, inthe RwembaitaSwamp, P. subfurcatum is the dominantlarval zygopteran andhas been observed in river sites in not stream or the park (APODACA, 2003). Thelow dis- solved levels oxygen that we observed in the RwembaitaSwamp are not unusually low for other in the papyrus swamps or heavily vegetated swamps region (BEADLE, 1932; CHAPMAN Based distribution CARTER, 1955; et al., 1998, 2000, 2001). on patterns of temperatezygoptems, this would represent stressful conditions for many temperate species (LAWTON, 1971; ERIKSEN, 1986). However, given the widespread nature of hypoxia in tropical fresh waters, it is possible that many aquatic insect groups have evolved such habitats.For larvae to exploit example, some zygopteran occupy the oxy- gen-scarce waters found in tree holes in tropical regions (DE LA ROSA & RAMIREZ, Thus, the to and/or deal with low 1995). ability physiologically behaviorally oxygen conditionsmay be pronounced in a suite of tropical damselfly larvae. SEASONAL PATTERNS OF OCCURRENCE Early investigators (KUMAR, 1976;CORBET, 1964; GAMBLES, 1960) suggested thatrainfallpatterns are extremely important in determining the seasonal distributionof odonates in seasonal In results that some highly tropical areas. general, our suggest sea- sonalfluctuations inrainfallhave littleeffect on the flying seasonofadult damselfliesin Kibale NationalPark. Pseudagrion spernatumwas the only species with a flight season that correlated with rainfall pattern. It was more likely to be observed in the dry sea- and found several months in both son was across swamp sites. Thus, its probability of dissolved low. If reflects occurrence was highest was oxygen was adult activity larval distribution, we anticipate high tolerance to hypoxia in the larvae of this species. Both Ceragrion glabrum andAfricallagma pseudelongatum were most active in the dry sea- 142 C.K. Apodaca & L.J. Chapman son and also foundin the RwembaitaSwamp. However, oursmall sample size forthese species and for Chlorocnemismarshallisuperba did not allow us to statistically test for significant predictors of occurrence. PARR (1984) found seven categories of flight ac- tivity for odonates in LiwondeNational Park, Malawi. With a greater sample size, our data for some species may fit a similarpattern; however, in general, season was not a significant predictor of occurrence. In where rainfallis distributed the areas more evenly throughout year, organisms may show a less pronounced response to seasonal change. Among tropical regions, there is variation in the and of the At lati- great onset, duration, magnitude wet season. high tudes, dry seasons may be severe and long, and produce strong seasonal variation in the availability and quality of aquatic habitats (e.g., northern Costa Rica, CHAPMAN & KRAMER, 1991), whereas in some equatorial regions, biannual rains or more continu- ous rainfall can buffer strong seasonal effects (e.g., central Congo basin, CHAPMAN & CHAPMAN, KUMAR annualrainfall of 2003). (1976) reported a single peak about 450 to 580 mm between August and mid-September in the Dumar Valley of India, where he investigated seasonal regulation of Zygoptera lifecycles. In this region, most adult activity, including breeding, takes place eitherbefore or after the wet season. In contrast, many damselfliesin Kibale National Park are active throughout the year. This reflect ofbiannual rainfall may the pattern that creates more even dispersion ofrainfall throughout the year. Clearly, there are seasonal cycles in limnological characters (e.g., dissolved at the sites, this and CHAPMAN et al., 1998, oxygen swamp study 2000), but they are not as pronounced as in some other tropical regions; and this may produce more seasonally consistent habitat availability for the zygopteran fauna. Our data suggest unregulated (not limitedby season or habitatavailability, essentially year-round) larval development in some species in Kibale National Park. Examples of in in unregulated development are most common coenagrionids lowland tropical ar- 1999). of eas (CORBET, The year round presence adult Proischnura subfurcatum and that these like othermembersoftheir Psuedagrion hageni suggest species, family, may followan unregulated emergencepattern. However, until we know more about emigra- tion andthe duration of the adult and larval periods for these species, it will be difficult to understandregulatory patterns governing their development. Lab-rearing experiments also will be required to fully understandemergence patterns, since the larval population is several classes represented by overlapping age (CORBET, 1999). CONSTRAINTS OF THE SAMPLING PROTOCOL The numberofspecies observed during our surveys did not include all thoseknown to occur in Kibale National Park. Two previously reported species, Pseudagrion kersteni and Chlorocnemispauli (D1JKSTRA & DINGEMANSE,2000), were not detected in our study. This may indicate that our survey method was not fully effective in detect- ing the seasonal presence of less common species, or that these species are extremely rare with periodic, perhaps supra-annual occupation of the habitats. Zygoptera ofKibale National Park, Uganda 143 When surveying for adult Zygoptera it is important totake their thermalrange ofactiv- into since ambient affect ofadult odonates ity account, temperature can activity patterns (MAY, 1977). Even if a species is present, it may not be active during all times of the day (LUTZ & PITTMAN, 1970). We tried to survey during the warmest times ofthe day, however, surveying in the rain was sometimes unavoidable. Thus, not all surveys were conducted underoptimal environmental conditions for adult activity. males. These results consist- Most individuals captured during our surveys were are with of odonatemales females ent previous reports outnumbering at water (CORBET, 1999; CORBET, 1962; JOHNSON, 1963; WAAGE, 1980), but this large discrepancy probably does not reflect the actual sex ratio of the adult population. It is more likely an artifact of the sampling protocol. Male behavior is more conspicuous, and surveys of females conducted at the waterside tend to detect a greater proportion males than (MICH1ELS& DHONDT, 1989). In fact, some investigators (BICK & BICK, 1968) biased ratios of 75% in This is in have reported male sex up to damselfly populations. differences in schedules and habitat between the part due to activity preferences sexes. Females visit the waterside as they become ready to breed, and males generally com- pete forterritories near water where they waitfor receptive females. Additionally, males and femalesof some species are active during different timesofthe day (MICHIELS & bias of DHONDT, 1989). Therefore, searching only at water can interpretations flight activity, but shouldbe sufficient to detect which species occupy a site. ACKNOWLEDGEMENTS the Wildlife The Center for Funding for this research was provided by Conservation Society, Sigma Xi, African Studies (University ofFlorida), and the Ford Foundation. We thank JAMES MAGARO, JOVAN MUGENYI, and MARGARET KEMIGISA of the KibaleFish and Monkey Project for their participation in collections. We thank VIOLA for ofher East field CLAUSNITZER use unpublishedkey to African odo- nates and K.-D.B. DUKSTRA for assistance with identifying damselfly specimens. Permission to conduct research in Uganda was acquired from the National Council for Science andTechnology and the Makerere University Biological Field Station ofUganda. Wealso thank FRANK NORDLIE, JAMESNATION, DAV- ID JULIAN, and NAT SEAVY for comments on an earlier version ofthis manuscript. REFERENCES ALLAN, D.J., 1995. Stream ecology: structure andfunctionofrunning waters. Chapman & Hall, London. APODACA, C.K., 2003.Damselflies in extreme environments: distribution andecophysiologyofProischnu- ra subfurcatum. M.S. thesis, Univ. Florida, Gainesville, Florida. ZISCHKE & G.L. 1982. Effect of elevated ARTFIUR, J.W., J.A. ERICKSEN, water temperature on macro- invertebrate communities in outdoor experimentalchannels. Water Res. 16: 1465-1477. BEADLE, L.C., 1932. Scientific results ofthe CambridgeExpedition to the East African Lakes, 1930-1. 3. Observations the bionomics of East African J. Linn. Soc. 38: 135-155. on some swamps. (Zool.) 1995. Lifetime and in the BENNETT, S. & P.J. MILL, egg production egg mortality damselfly Pyrrhosoma nymphula (Sulzer) (Zygoptera: Coenagrionidae). Hydrobiologia 310:71-78. BICK, G.H. & J.C. BICK, 1968. Demography ofthe damselfly, Argia plana Calvert (Odonata: Coenagrii- dae). Proc. ent. Soc. Wash. 70: 197-203. 144 C.K, Apodaca & LJ. Chapman C.A. BORSUK, M.E., STOW, R.A. LUETTICH, H.W. PAERL & J.L. PINCKNEY, 2001. Modellingoxy- in stratified estimation of field data. gen dynamics an intermittently estuary: process rates using Es- tuarine coastal ShelfSei. 52: 33-49. BUTYNSKI, T.M.L., 1990, Comparative ecology ofblue monkeys (Cercopithicus mitis) in high and low- density subpopulations. Ecol. Monogr. 60: 1-26. 1987. Odonata of saline lakes in British Can- CANNINGS, R.A. & S.G. CANNINGS, The some Columbia, ada: ecological distribution and zoogeography.Adv. Odonatol. 3: 7-21. 1955. The CARTER, G.S., papyrus swamps of Uganda. Heffer,Cambridge. CHAPMAN, C.A. & J.E. LAMBERT, 2000. Habitat alterations and the conservation ofAfrican primates: of case study Kibale National Park, Uganda. Am. J. Primatol. 50: 169-185. CHAPMAN, L.J., J. BALIRWA, F.W.B. BUGENYI,C.A. CHAPMAN & T.L. CRISMAN,2001. Wetlands ofEast Africa: biodiversity,exploitation, and policy perspectives. In: B. Gopal, [Ed,], Wetlands bio- diversity, pp. 101-132, Backhuys, Leiden. CHAPMAN, L.J. & C.A. CHAPMAN, 2003. Fishes of the African rain forests: emerging and potential threats to a little-known fauna, ln: T.L. Crisman, L.J. Chapman, C.A. Chapman,& L.S. Kaufman. and Florida [Eds], Conservation, ecology, management ofAfricanfreshwaters, pp 176-209. Univ. Press, Gainesville/FL. — [In press]. CHAPMAN, L.J., C.A. CHAPMAN & T.L. CRISMAN, 1998. Limnological observations of a papyrus and Verh. swamp in Uganda: implications for fish faunal structure diversity. int. Ver. Umnol. 26: 1821-1826. CHAPMAN, L.J., C.A. T.L. CRISMAN & J. 2000. Predictors ofseasonal CHAPMAN, PRENGER, oxygen levels in a Ugandan swamp/riversystem; a 3-year profile. Verh. int. Ver. Limnol. 27: 3048-3053. & CHAPMAN, L.J. D.L. KRAMER, 1991. Limnologicalobservations ofanintermittent tropical dryforest stream. Hydrobiologia226: 153-166. CLAUSNITZER, V., 2001. Notes on the species diversity ofEast African Odonata, with a checklist of spe- cies. Odonatologica30: 49-66. 1955. A of in Nature CORBET, P.S., critical response to changinglength day an insect. (Land.) 175: 338- -339. CORBET, P.S., 1962. Abiology of dragonflies.Witherby, London. CORBET, P.S., 1964. Temporal pattern ofemergence in aquatic insects. Can. Ent. 96: 264-279. CORBET, P.S., 1980. Biology ofOdonata. Annu. Rev. Ent. 25: 189-217. CORBET, P.S., 1999. Dragonflies: behavior and ecologyof Odonata. Cornell Univ. Press, New York. L.A. & W.H. 1998. Effects ofacidic COURTNEY, CLEMENTS, pH on benthic macroinvertebrate com- munities in stream microcosms. Hydrobiologia379: 135-145. DAVIS, J.C., 1975. Minimal oxygen requirements of aquatic life with emphasis on Canadian species: a re- view. J. Fish. Res. Bd Can. 32: 2295-2332. DE LA ROSA, C. & A. RAMIREZ, 1995. A note on phototactic behavior and on phoretic associations in larvae ofMecistogaster ornata Rambur from northern Costa Rica (Zygoptera: Pseudostigmatidae). Odonatologica 24: 219-224. DUKSTRA, K.-D.B. & N.J. DINGEMANSE, 2000. Odonata from Kibale National Park, western Uganda. Notul. odonatol. 5: 72-75. caudal lamellae in ERIKSEN, C.H., 1986. Respiratory roles ofthe (gills) a lestid damselfly (Odonata: Zy- goptera). Jl N. Am. benthol. Soc. 5: 16-27. ERWIN, T.L., 1982. Tropical forests: their richness in Coleoptera and other arthropod species, Coleopts Bull. 36: 74-75, small FEMINELLA, J.W., 1996. Comparison ofbenthic macroinvertebrate assemblages in streams along a offlow N. Am. benthol. Soc. 15: gradient permanence. Jl 651-669. FENCHEL, T. & R.N. 2000. Benthic GLUD, primary production and 0 C0 dynamics in a shallow-water 2 2 sediment: spatialand temporal heterogeneity.Ophelia 53: 159-171. FLEMER, D.A., W.L. KRUCZYNSKI, B.F. RUTH & C.M. BUNDRICK, 1999. The relativeinfluence of Zygoptcra of Kibale National Park, Uganda 145 hypoxia, anoxia, and associated environmental factors as determinants ofmacrobenthic community J. Recov. structure in a Northern Gulf ofMexico estuary. aquat. Ecosyst. Stress 6: 311-328. GAMBLES, R.M., 1960. Seasonal distribution and longevity in Nigerian dragonflies. J. W. Afr. sci. Ass. 6: 18-26. GARRISON, R.W. & J. MUZON, 1995. Collecting down at the other “down under.” Argia 7(3): 23-26. size GILLOOLY, J.F. & S.I. DODOSON, 2000. The relationship between egg and incubation tempera- ture to embryonic development time in univoltine and multivoltine aquatic insects, Freshw. Biol. 44: 595-604. HALL, R.J. & F.P. IDE, 1987. Evidence of acidification effects on stream insect communities in central On- tario between 1937 and 1985. Can. J. Fish, aquat. Sei. 44: 1652-1657. HALVERSON, T.G., 1983. Temperature dependent embryogenesis in Aeshna tuberculifera Walker and Plathemis lydia (Drury) under field and laboratoryconditions (Anisoptera: Aeshnidae, Libellulidae). Odonatologica 12: 367-J73. HAWKING, J.H. & T.R. NEW, 1995. Development ofeggs of dragonflies (Odonata: Anisoptera) from two streams in north-eastern Victoria,Australia. Aquat. Insects 17: 175-180. HUBERTZ, E.D. & L.B. CAHOON, 1999. Short-term variability ofwater quality parameters in two shallow estuaries ofNorth Carolina. Estuaries 22: 814-823. H. WALTHER & R. 1984. insect larvae indicators of minimal JACOB, U., KLENKE, Aquatic as limiting content of dissolved oxygen. 2. Aquat. Insects 6: 185-190. JOHNSON, C„ 1963. Breeding structure in populationsof the Odonata. Texas J. Sei. 15: 171-183. KELTS, L.J., 1979. Ecology of a tidal marsh corixid, Trichocorixa verticalis (Insecta, Hemiptera). Hydro- biologia 64: 37-57, KUMAR, A., 1972. The phenology of dragonfliesin the Dehra Dun Valley, India. Odonatologica 1; 199- -207. KUMAR, A., 1976. Biology ofIndian dragonflieswith special reference to seasonal regulation and larval development.Bull. Ent. 17: 37-47. LAHR, J., A.O. DIALLO, K.B. NDOUR, A. BADJI & PS. DIOUF, 1999. Phenology ofinvertebrates liv- ing in asahelian temporary pond. Hydrobiologie 405: 189-205. LAWTON, J.H., 1970. A populationstudy on larvae ofthe damselfly Pyrrhosomanymphula (Sulzer) (Odo- nata; Zygoptera). Hydrobiologie36: 33-52. LAWTON, J.H., 1971. Ecologicalenergetics studies on larvae ofthe damselfly Pyrrhosoma nymphula(Sulz- er) (Odonata; Zygoptera). J. Anim. Ecol. 40: 385-423. A. in LUTZ, P.E. & ROGERS, 1991. Thermal effects onembryonic development four summer species of Libellulidae (Anisoptera). Odonatologica 20: 281-292. LUTZ, P.E. & A.R., PITTMAN, 1970. Some ecological factors influencing a community of adult Odonata. Ecology 51:279-284. 1977, and in of the Micra- MAY, M.L., Thermoregulation reproductive activity tropical dragonflies genus thyria. Ecology 58: 787-798. MERRITT, R.W. & K.W. CUMMINS, [Eds], 1996.An introductionto the aquatic insects ofNorth America, [3rd edn], Kendall/Hunt, Dubuque, LA. MICHIELS, N.K. & A.A. DHONDT, 1989. Differences in male and female activity patterns in the dragon- fly Sympetrum danae(Sulzer) and their relationto mate-finding(Anisoptera:Libellulidae).Odona- tologica 18: 349-364. NORLING, U., 1975.Livscykler hos svenska odonater. Entomologen4: 1-14. NORUSIS, M.J./SPSS Inc., 1999. Regression models™ 10.0. SPSS Inc., Chicago/IL. PARR, M.J., 1984. The seasonal occurrence of Odonata in the Liwonde Park, Malawi. Adv. Odonatol. 2: 157-167. The and larval in Odonata. Can. PROCTER, D.L., 1973. effectoftemperature photoperiodon development J.Zool. 51: 1165-1170. ROUSCH, J.M., T.W. SIMMONS, B.L. KERANS & B.P, SMITH, 1997. Relative acute effects of low pH 146 C.K. Apodaca & L.J. Chapman iron the and the the and high on hatching survival of water mite (Arrenurus manubriator)and aquatic insect (Chironomusriparius). Envir. Toxicol. Chem. 16: 2144-2150. UBUKATA, H., 1980. Life history and behavior ofa corduliid dragonfly, Cordulia aeneaamurensis Selys. 3. Aquatic period, with special reference to larval growth. Kontyu48: 414-427. WAAGE, J.K, 1980. Adult sex ratios and female reproductive potential in Calopteryx (Zygoptera: Calop- terygidae). Odonatologica9: 217-230. WELLKITZ, T.A., NX. POFF, G. COSYLEON & B. STEURY, 2001. Current velocity and spatial scale ofthe distribution and abundanceof herbivorous as determinants two rheophilic insects. Landscape Ecol. 16: 111-120. WIGGLESWORTH, V.B., 1972. The principles ofinsect physiology. [7th edn], Chapman & Hall, London. WILLIAMS, F.X., 1936. Biological studies in Hawaiian water-lovinginsects, 2. Odonata or dragonflies. Proc. Hawaii, ent. Soc. 9: 273-349. WINSTANLEY, W.J., 1983. Terrestrial larvae of Odonata from New Caledonia (Zygoptera; Megapoda- grionidae; Anisoptera: Synthemistidae),Odonatologica 12: 389-395. WOLANSKI, E. & E. GERETA, 2001. Water quantity and quality as the factors driving the Serengeti eco- system, Tanzania. Hydrobiologia458: 169-180.