Tefferi 2014_Drying Effects Tropical Reservoirs

Strong effects of occasional drying on subsequent waterclarity and cyanobacterial blooms in cool tropical reservoirs

MEKONEN TEFERI* , † , STEVEN A. J . DECLERCK‡, TOM DE BIE* , PIETER LEMMENS*, ABRAHA

GEBREKIDAN§ , ¶ , TSEHAYE ASMELASH** , TADESSE DEJENIE† , KINDEYA GEBREHIWOT†† ,

HANS BAUER‡‡ , JOZEF A. DECKERS§§ , JOS SNOEKS ¶ ¶ , * * * AND LUC DE MEESTER*

*Laboratory of Aquatic Ecology, Evolution and Conservation, KU Leuven, Leuven, Belgium†Department of Biology, Mekelle University, Mekelle, Ethiopia‡Department of Aquatic Ecology, Netherlands Institute of Ecology (NIOO-KNAW), Wageningen, the Netherlands§Department of Chemistry, Mekelle University, Mekelle, Ethiopia¶Department of Chemical Engineering, Process Engineering for Sustainable Systems, KU Leuven, Heverlee, Belgium

**Department of Microbiology, Mekelle University, Mekelle, Ethiopia††Department of Land Resource Management and Environmental Protection, Mekelle University, Mekelle, Ethiopia‡‡WildCRU, University of Oxford, Tubney House, U.K.§§Division of Soil and Water Management, KU Leuven, Leuven, Belgium¶¶Laboratory of Biodiversity and Evolutionary Genomics, KU Leuven, Leuven, Belgium

***Department of Biology, Royal Museum for Central Africa, Tervuren, Belgium

SUMMARY

1. In semi-arid regions, the construction of small reservoirs is important in alleviating water shortage,

although many have poor water quality with high turbidity and dense blooms of algae and cyano-

bacteria, and there are large differences in the ecology of such reservoirs.

2. We took advantage of two exceptionally dry years in northern Ethiopia to study the effect of a dry

period and the associated fish kills on reservoir ecology and water quality. We studied 13 reservoirs,

seven of which dried up in 2009. Four of the latter dried up again in 2010. We monitored the ecology

of these reservoirs from 2009 to 2011, hypothesising that the pattern of reservoir drying would

explain ecological differences among them.

3. Reservoirs that refilled after drying had a significantly lower fish biomass, lower biomass of

phytoplankton (expressed as chlorophyll-a) and cyanobacteria (Microcystis), clearer water, greater

macrophyte cover and lower nutrient concentrations than reservoirs that did not dry. Although the

differences in water quality were most striking in the wet season after a drying event, there were

persistent effects on reservoir ecology. The three categories of reservoirs we distinguished, based on

their behaviour in 2009 and 2010, also showed differences in 2004, a year during which none of the

reservoirs dried out. While drying evidently results in better water quality, we could not disentangle

the effects of drying per se from that of reductions in fish biomass. The total combined effect was

highly significant in all 3 years, whereas the separate effects of drying and loss of fish were only

significant in 2004.

4. Our results suggest that differences in water quality and ecology among reservoirs depend on

their propensity to dry out. Drying might be used as a restoration measure to reduce potentially

harmful cyanobacterial blooms in reservoirs.

Keywords: Ethiopia, fish biomass, Microcystis, reservoir ecology, water clarity

Correspondence: Mekonen Teferi, Laboratory of Aquatic Ecology, Evolution and Conservation, KU Leuven, Charles De Beriotstraat 32, Bus

2439, B-3000 Leuven, Belgium and Department of Biology, Mekelle University, Mekelle, Ethiopia. E-mails: [email protected], senaymt@

yahoo.com

870 © 2014 John Wiley & Sons Ltd

Freshwater Biology (2014) 59, 870–884 doi:10.1111/fwb.12312

Introduction

Reservoirs are large, man-made impoundments of water,

usually constructed where natural freshwater bodies are

rare or where the water available is unsuitable for human

use (Lowe-McConnell, 1966; Wetzel, 2001). These water-

bodies are constructed worldwide for a variety of

purposes, mostly for irrigation, flood control, household

use, watering of livestock or hydropower. Most reservoirs

in semi-arid regions are shallow and show wide fluctua-

tions in water level (Gasse, 2000; Vallet-Coulomb et al.,

2001; Coops, Beklioglu & Crisman, 2003; Naselli-Flores &

Barone, 2005; Beklioglu, Altinayar & Tan, 2006; €Ozen

et al., 2010; Bucak et al., 2012). Seasonal fluctuations in

water level are generally larger in reservoirs than in

natural lakes. Despite their importance, many reservoirs

are poorly managed and suffer from a wide range of

problems, such as excessive algal blooms, toxic cyanobac-

teria, turbidity, exotic species and increased salinity

(Holdren, Jones & Taggart, 2001). Toxic cyanobacterial

blooms in eutrophic waters (Codd, Ward & Bell, 1997;

Chorus & Bertram, 1999; Graham, 2007) can cause health

hazards to animals and humans (Bell & Codd, 1994; Codd

et al., 1997; Codd, Metcalf & Beattie, 1999). Understanding

the mechanisms that determine water quality and the

occurrence of cyanobacterial blooms in reservoirs is,

therefore, of crucial importance for management.

Much research has been carried out on the restoration

of eutrophic lakes (Scheffer, 1998; Moss, 2010). Reduc-

tion in external nutrient inputs is often not sufficient to

restore water clarity in shallow lakes and reservoirs.

Additional measures are needed, such as biomanipula-

tion of the fish assemblage, removal of sediments or lake

drawdown (Jeppesen et al., 2005, 2012; Van Wichelen

et al., 2007). Several studies in temperate regions have

shown that complete drying of waterbodies may be a

powerful management tool in shallow lakes (Van Geest

et al., 2005). A complete drawdown results in a consoli-

dation of the lake sediment and nutrient loss due to per-

manent sedimentation (James et al., 2001; Søndergaard,

Jensen & Jeppesen, 2003). A complete drawdown also

facilitates control of fish biomass and removal of accu-

mulated sediments (Scheffer, 1998; Jeppesen et al., 2012).

Benthivorous fish are a key cause of turbidity in shallow

lakes as they stir up the sediment during feeding

(Havens, 1993; Horppila et al., 1998; Scheffer, Portielje &

Zambrano, 2003). In contrast to the large body of litera-

ture on lake management in temperate regions, very

little is known about how reservoir drying affects the

ecology of reservoirs in the tropics and subtropics

(Jeppesen et al., 2012).

In the semi-arid highland region of Tigray, North

Ethiopia, more than 70 small reservoirs with a surface

area between two and 50 hectares have been constructed

over the last three decades (Haregeweyn et al., 2006;

Dejenie et al., 2008). Due to severe soil erosion caused by

land degradation (Nyssen et al., 2005), excessive nutrient

loads are associated with sediment influx to these reser-

voirs (De Wit, 2003; Haregeweyn et al., 2006). Thus,

Dejenie et al. (2008) surveyed 32 reservoirs in Tigray and

found that many had high nutrient concentrations, high

phytoplankton biomass, low water transparency and no

or only a poorly developed vegetation of submerged

macrophytes. Cyanobacteria blooms (mainly Microcystis)

are frequent (Dejenie et al., 2008; Van Gremberghe et al.,

2011) in these reservoirs. Daphnia are generally rare in

the tropics (Chiambeng & Dumont, 2005), but they do

occur in fishless lakes (Iglesias et al., 2011), at higher alti-

tudes (Mergeay et al., 2006), and they also occur in the

reservoirs of the Ethiopian highlands (Dejenie et al.,

2008, 2012). With the exception of introduced Tilapia

(Oreochromis niloticus or Tilapia zillii) in a few locations,

fish assemblages in the northern Ethiopian reservoirs are

composed exclusively of species of the riverine Cyprinid

Garra. This genus is widespread in rivers and lakes of

Asia and Africa (Getahun & Stiassny, 1998; Tudorancea

& Taylor, 2002; Zhou, Pan & Kottelat, 2005). Large pop-

ulations of Garra are found in many reservoirs in Tigray,

probably as the result of high nutrient loading and the

absence of piscivorous fish (Dejenie et al., 2008; Teferi

et al., 2013). Many tropical and warm subtropical lake

systems are indeed characterised by a very high abun-

dance of small fish (Meerhoff et al., 2007; Teixeira-de

Mello et al., 2009).

High densities of benthivorous fish can increase nutri-

ent availability and lead to phytoplankton blooms

through excretion and resuspension of sediments

(Scheffer, 1998; Teixeira-de Mello et al., 2009; Jeppesen

et al., 2012). Indeed, Dejenie et al. (2008) found that the

biomass of Garra was associated with high concentration

of phosphorus in Ethiopian reservoirs. In enclosures with

Garra, they also found more nutrients and higher biomass

of phytoplankton and zooplankton in the water column,

and higher amounts of suspended matter than in fishless

enclosures (Dejenie et al., 2009), suggesting that Garra

may increase primary production indirectly through

bottom-up effects. Garra could also increase phytoplank-

ton biomass by exerting a top-down effect on Daphnia

(Dejenie et al., 2009).

In this study, we took advantage of a natural whole-

reservoir experiment caused by two consecutive excep-

tionally dry years in Tigray. A poor rainy season in

© 2014 John Wiley & Sons Ltd, Freshwater Biology, 59, 870–884

Impact of reservoir drying on water quality 871

2008, followed by an extended dry period, resulted in a

substantial number of reservoirs in the highlands of Ti-

gray drying completely in April–June 2009, and some

went dry again in April–June 2010. We took this oppor-

tunity to test for the effects of a reservoir drying out for

a limited period, and the associated fish kills, by com-

paring temporal ecological changes between reservoirs

that dried out and those that remained wet in April–

June 2009 and 2010. We hypothesised that reservoir dry-

ing would improve water quality and reduce cyanobac-

terial blooms in tropical reservoirs.

Methods

Study area and design

Tigray is in the northern Ethiopian highlands between

12° and 15°N and between 37°10′E and 40°10′E (see

Figure S1 in Supporting Information). It belongs to the

African dry lands, often referred to as Sudano–Sahelian

region (Warren & Khogali, 1992). The climate is cool

tropical continental, with an extended dry season of nine

to 10 months and a maximum effective rainy season of

50–60 days in July–August (HTS, 1976). Average annual

rainfall in Tigray is between 450 and 980 mm (data from

National Meteorology office, Mekelle branch; Lemma,

1996).

We selected 13 reservoirs that were a subset of the 32

previously studied by Dejenie et al. (2008) and represent

both reservoirs that did and did not dry up in 2009 and

2010, two exceptionally dry years. We distinguished

three categories: reservoirs that dried up in both 2009

and 2010, reservoirs that dried up only in 2009 and res-

ervoirs that did not dry up during the study period

(Fig. 1; Table 1). All reservoirs studied were visited reg-

ularly in the period 2003–2011, and none dried in the

period from 2003 till 2008. All reservoirs lose water due

to evaporation; the fact that some dried up completely

and others did not was in part due to differences in

water withdrawal for irrigation purposes.

Sample collection and processing

All reservoirs were sampled in September 2009, 2010

and 2011, when the reservoirs were at full capacity.

Sampling took place between 10 and 25 September, that

is, just after the rainy season, and between 9:00 a.m. and

12:00 a.m. On each sampling occasion and for each res-

ervoir, we measured a range of morphometrical, regio-

nal and local ecological characteristics. The

morphometric and regional characteristics measured

were surface area, depth, altitude and reservoir age.

Abiotic environmental variables measured were Secchi

depth, pH, oxygen, nutrient concentrations (total phos-

phorus and total nitrogen), suspended matter, conduc-

tivity and temperature. In addition, we quantified a

number of biotic variables related to regime shifts

between the turbid and the clear-water state in shallow

lakes (Scheffer et al., 2003): vegetation cover, fish bio-

mass, phytoplankton biomass (as chlorophyll-a concen-

Fig. 1 Fish biomass (in three categories:

high fish biomass/low fish biomass/no

fish) through time in the 13 reservoirs

studied indicating time of fish recolonisa-

tion after reservoir drying. High fish

biomass means that the biomass caught

in gillnets was >10% of average of all

reservoirs; low fish biomass <10% of

average of all reservoirs. Fish were

sampled in all reservoirs by gillnetting

four times a year (March, August,

September and December).


872 M. Teferi et al.

tration), Microcystis biomass, zooplankton biomass and

Daphnia biomass.

Depth was quantified as the mean of the depths at

two pelagic sampling stations that were known to be in

the deeper portion of the reservoirs. Surface area was

calculated from multiple GPS recordings along the reser-

voir shores. Reservoir age and dominant water use were

retrieved from SAERT (1994), interactions with local

communities and our own observations, and mainly

included irrigation, watering of livestock and water use

for household activities. In principle, the reservoirs are

not created for supplying drinking water to people, but

regularly we observed people drinking from the reser-

voirs as they were working in the fields or herding cattle

(M. Teferi and A. Gebrekidan, pers. obs.).

On each sampling occasion, depth-integrated water

samples were taken at two pelagic sites with a Heart

valve sampler (volume: 3 L). At each site, we took water

samples at three depths (surface, mid-water and near

the bottom). After pooling the water samples, subsam-

ples were taken for the analysis of cyanobacteria, chloro-

phyll-a concentration, total phosphorus (TP), total

nitrogen (TN) and suspended matter. We used cali-

brated fluorometer readings (Turner Aquafluor; mean of

three measurements) on the pooled water sample to

measure chlorophyll-a concentration as a proxy of phy-

toplankton biomass. We determined suspended matter

as the dry mass of particles retained on pre-weighed

Whatman GF/C filter divided by the volume of water

filtered. The total phosphorus and total nitrogen concen-

trations were measured following the ascorbic acid and

Kjeldahl methods, respectively (APHA, 1999). Dissolved

oxygen, pH, conductivity and temperature were mea-

sured at two pelagic stations with a WTW Multi 340 I

electrode. Secchi depth was also measured at the same

two stations (disc diameter: 30 cm). Zooplankton sam-

ples were collected with a Schindler–Patalas plankton

trap of 12 L volume (64-lm mesh size) at the two cen-

trally located pelagic sites and then pooled. At each site,

we sampled at 1-m-depth intervals from the surface to

the bottom. Samples were fixed with sucrose-saturated

formalin solution (4% final concentration). Cladocerans

were identified to species using keys by Flossner (2000)

initially and then referring to Smirnov (1992, 1996).

Biomass estimates were obtained using published

length–mass regressions (Dumont, Velde & Dumont,

1975; Bottrell et al., 1976). Cyanobacteria were identified

from acid Lugol-fixed samples following Whitton, Brook

and John (2002), Whitford and Schumacher (1973) and

Komarek and Anagnostidis (2000, 2005). Cyanobacteria

were quantified from counts of depth-integrated samples

and converted to biomass following published conver-

sions (Hillebrand et al., 1999).

Fish were sampled repeatedly in all 13 reservoirs in

the period 2009–2010 (August 2009, September 2009,

December 2009, March 2010, August 2010, September

2010 and December 2010) to monitor recolonisation by

Garra of the formerly dry reservoirs when they had

refilled. Fish were sampled by deploying eight standar-

dised multimesh gillnets overnight, with four gillnets in

the pelagic and four gillnets in the littoral zone of each

reservoir (Appelberg, 2000). The nets were deployed in

the late afternoon and retrieved the following morning,

after about 12 h. Fish biomass was estimated by direct

weighing in the field. Data represent mean biomass

among nets, expressed as catch per unit effort (CPUE; g

Table 1 Drying history and classification of the

studied reservoirs. N represents reservoirs that

did not dry up at the end of the dry season (April

–June) of a given year; D represents reservoirs that

dried up completely in the given year. Reservoirs

are allocated to one of three categories based on

whether they dried up in 2009 and 2010

Reservoir

name

Coordinates Drying

history

April–June2009

Drying

history

April–June2010 CategoriesLongitude Latitude

Adi Asmee 38.96 13.65 N N ‘Permanent in study

period’ (n = 6)Bokoro 39.57 14.20 N N

Era Quihila 39.60 13.45 N N

Gereb Awso 39.56 13.43 N N

Ruba Feleg 39.73 13.95 N N

Tsinkanet 39.54 14.01 N N

Dur Anbesa 39.44 13.27 D N ‘Only dry in 2009’

(n = 3)Gereb Beati 39.48 13.45 D N

Shilanat IV 39.49 13.10 D N

Gereb Mihiz 39.47 13.29 D D ‘Dry in 2009 and

2010’ (n = 4)Gum Selasa 39.54 13.24 D D

Mai Gassa I 39.49 13.29 D D

Mai Gassa II 39.49 13.29 D D



fresh mass per net per night). Each gillnet was 28 m

long, 3 m wide and was composed of seven frames of

4 m each with a different mesh size (6.25, 10.5, 15.5, 24,

31, 39 and 46 mm knot-to-knot).

In all reservoirs, the presence and abundance of mac-

rophytes were estimated as percentage cover (10% inter-

val) by inspecting both the littoral and pelagic parts of

the reservoir by boat. Spatial variables including lati-

tude, longitude and altitude were taken from Dejenie

et al. (2008).

Statistical analyses

To test whether drying influenced limnological and eco-

logical characteristics of the reservoirs, we took both a

multivariate and a univariate approach. First, we applied

standardised redundancy analysis (RDA) to the entire set

of measured variables to test for any systematic differenti-

ation between the three categories of reservoirs (Leps &

Smilauer, 2003; Legendre & Legendre, 2012). We used

standardised principal component analysis (PCA) to

explore visually patterns of associations among variables

and between variables and the drying categories. All these

analyses were carried out for each sampling year sepa-

rately to evaluate the consistency of patterns across years.

In addition to the multivariate analysis, we studied the

detailed response of each variable to the drying regimes

with one-way ANOVAs testing for an effect of drying

regime (never dry, sometimes dry and regularly dry) on

each of the variables separately.

In addition to the sampling campaigns of 2009, 2010

and 2011, we also analysed the data obtained for the same

13 reservoirs in September 2004. The 2004 data are a sub-

set of the data analysed by Dejenie et al. (2008). The sam-

pling protocols in this study were identical to those

applied in this study. Given that none of the reservoirs

dried out in the 2 years preceding September 2004, this

analysis provided us with a longer-term perspective on

the influence of previous drying on the ecology of the res-

ervoirs in periods when no such drying occurred. In addi-

tion, given that there was a 5- to 7-year time period

between this first and the current sampling campaigns,

the combined data also provided some insight into the

stability of the ecological characteristics of the reservoirs

as characterised by their propensity to dry out.

The effects of drying may result from the associated

fish kills that can potentially have strong effects on the

ecology of reservoirs, because of their role in the aquatic

food web and their impact on sediment resuspension

and internal eutrophication. Drying could, however, also

have effects independent of those of fish. We carried out

a variation partitioning analysis using partial RDA

(Peres-Neto et al., 2006) to decompose total variation in

environmental variables into a purely fish-related com-

ponent, a purely drying-related component, a compo-

nent representing the combined effects of both, and the

remaining unexplained variation.

Prior to analyses, we log-transformed all environmen-

tal variables except pH. RDAs were carried out with

CANOCO v4.5, and the significance of the models was

tested with 999 Monte Carlo permutations (reduced

model). Partial RDA for variation partitioning was per-

formed in R v2.8.1; R Development Core Team 2008,

using the RDA and VARPART functions of the vegan

library (Oksanen, 2005; Peres-Neto et al., 2006). ANOVA

and Tukey’s post hoc comparisons were performed using

the statistical software package STATISTICA 11 (StatSoft,

Inc., Tulsa, OK, U.S.A).

Results

Recolonisation of reservoirs by fish following drying

Reservoirs that dried out naturally became fishless. The

reservoirs refilled during July and August, mainly by

surface run-off. Of the seven reservoirs that dried during

the dry period in 2009, three were recolonised by fish

immediately upon refilling in August 2009 (Fig. 1). Of

the remaining four reservoirs, two had been recolonised

by December 2009 and one by March 2010. One reser-

voir (Gum Selasa) remained fishless for an entire year

(Fig. 1; see also Table 1). During the dry period of 2010,

four reservoirs dried up again for a second time, fish re-

colonising them within 1–2 months of their refilling,

except again for Gum Selasa, where we caught no fish

until December 2010.

Impact of drying on reservoir characteristics: 2009–2011

The three categories of reservoirs (dry in both 2009 and

2010; dry only in 2009; no drying) did not differ in any

of the morphometric and regional variables measured,

except for surface area (see Table S1; Figure S2). The res-

ervoirs that were more likely to dry were generally

large, because large reservoirs are used more intensively

for irrigation.

Drying had a profound impact on the ecology of the

reservoirs. RDA performed on the entire set of local envi-

ronmental conditions showed a strong effect in each of

the years studied. The proportion of total variation in

these variables explained by drying was 33.8%

(F-ratio = 2.558, P = 0.009) in 2009, 40.5% (F-ratio = 3.409,



P = 0.002) in 2010 and 37.9% (F-ratio = 3.048, P = 0.006)

in 2011 (Table S2). PCA also showed consistent

differences among the three reservoir categories across

years. PCA1 always showed high eigenvalues (39.5%,

40.7% and 42.4% in 2009, 2010 and 2011, respectively)

and in all 3 years reflected a gradient from drying

reservoirs with few or no fish, relatively high macrophyte

cover and high transparency to permanent reservoirs

with abundant fish, phytoplankton and Microcystis

biomass and high TP and suspended material (Fig. 2).

Zooplankton and Daphnia biomass rather tended to show

no or only a weak association with this gradient. From

the PCA, it is also clear that reservoirs that fell dry only

in 2009 showed intermediate characteristics to the reser-

voirs that did not dry out and the reservoirs that dried in

both years (Fig. 2).

Detailed ANOVAs followed by Tukey post hoc compari-

sons performed on individual key variables revealed that,

in each of the three consecutive years, permanent reser-

voirs had a significantly higher fish biomass, total phyto-

plankton biomass (chlorophyll-a), Microcystis biomass,

total phosphorus and suspended matter concentration

than reservoirs that dried out once (2009) or twice (2009

and 2010) (Fig. 3; Table 2). Although also pronounced,

patterns tended to be somewhat more complex and vari-

able through time for macrophyte cover and Secchi depth

(Fig. 3). Just after the first drying in 2009, macrophyte

cover and water transparency were substantially higher

-1.0 1.0

-1.0

1.0

Fish

Chl-a

Microcystis

Macrophyte

ZooplanktonDaphnia

TPTN

SM

Secchi

Depth

Oxygen

Temperature

pH

Conductivity

Area

Altitude

Age

AAs

BOK

EQ

GA

RF

TS

DAGB

SHIV

GM

GS

MGI

MGIINon dry 2009 & 2010

Dry 2009 Dry 2009 & 2010

2009

Axis1: Eigenvalue = 39.5%

Axi

s2: E

igen

valu

e =

16.5

%

-1.0 1.0

-1.0

1.0

Fish

Chl-a

MicrocystisMacrophyte

ZooplanktonDaphnia

TPTN

SM

SecchiDepth

Oxygen

Temperature

pHConductivity

Area

Altitude

Age

AAs

BOK

EQ

GA

RF

TS

DAGB

SHIV

GM

GS

MGI

MGII

Non dry 2009 & 2010

Dry 2009Dry 2009 & 2010

2010

Axis1:Eigenvalue = 40.7%

Axi

s2:E

igen

valu

e =

13.6

%

-1.0 1.0

-1.0

1.0

Fish

Chl-aMicrocystis

Macrophyte

ZooplanktonDaphnia

TP

TN SM

Secchi

Depth

Oxygen

Temperature

pH

Conductivity

Area

Altitude

Age

AAs

BOK

EQGA

RF

TS

DA

GBSHIV

GM

GS

MGI

MGII

Non dry 2009 & 2010Dry 2009

Dry 2009 & 2010

2011


Axi

s2:E

igen

valu

e =

16.6

%

-1.0 1.0

-1.0

1.0

Fish

Chl-a

Microcystis

Macrophyte

Zooplankton

Daphnia

TP

TN

SM

Secchi

Depth

Oxygen

Temperature

pH

Conductivity

Area

Altitude

Age

AAs

BOK

EQGA

RF

TS

DA

GB

SHIV

GM

GS

MGI

MGII

Non dry 2009 & 2010

Dry 2009

Dry 2009 & 2010

2004


Axi

s2:E

igen

valu

e =

20.8

%

Fig. 2 Triplot of standardised PCA on the data for environmental variables of reservoirs that did not dry out in 2009–2010 (filled circles),

dried only in 2009 (filled squares) or dried in both 2009 and 2010 (filled diamonds), separately for the sampling campaigns of September

2009, 2010 and 2011, and for the data of 2004 (Dejenie et al., 2008). Reservoir names are given in abbreviations: for example, Adi Asme’e:

AAs; Bokoro: BOK (for full names of reservoirs, see Table 1). Filled down-pointing triangles represent centroids and show the average posi-

tion of each category of reservoirs.



in the reservoirs that had fallen dry than in the permanent

systems. During the year of the second drought (2010)

and in the subsequent year (2011), however, a clear differ-

entiation occurred between reservoirs that had only fallen

dry once and reservoirs that fell dry twice consecutively,

the latter being characterised by much higher macrophyte

cover and water transparency, whereas the former shifted

back to a situation similar to permanent reservoirs (Fig. 3;

Table 2). The reservoirs that dried out only in 2009 gener-

ally showed intermediate characteristics to those that fell

Fish

bio

mas

s (g)

0

5000

10 000

15 000

20 000

25 000

30 000Permanent Dry 2009Dry 2009 & 2010

a

bc

a

b c

a

b b

a

bb

Zoo

plan

kton

bio

mas

s (µg

L–1

)

0

200

400

600

800

1000 a

bb

Dap

hnia

bio

mas

s (µg

L–1

)

0

10

20

30

40

a

Tot

al P

hosp

horu

s (µg

L–1

)

0

100

200

300

400

500

600

a

abb

a

a

b a

ab b

Tot

al N

itrog

en (

µg L

–1)

0

2000

4000

6000

8000

Susp

ende

d m

atte

r (m

gL–1

)

0

50

100

150

200

250

300 a

b b

a

ab

b

a

ab

ab bb a

Secc

hi d

isk

tran

spar

ency

(m)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

a

b

b

aab

ba a

b

Tem

pera

ture

(0 C)

0

5

10

15

20

25

30

201120102009 2004

20112009 20042010 20112009 20042010

Dis

solv

ed o

xyge

n (m

g L

–1)

0

2

4

6

8

10

12

14

aa

b

aab b

pH

0

2

4

6

8

10

12

14

Con

duct

ivity

(µS

cm–1

)

0

100

200

300

400

500

Chl

orop

hyll

a (µ

g L

–1)

0

200

400

600

800 a

b

a

bb

a

b bb a

Microcystis b

iom

ass (

µg L

–1:L

og)

0

2

4

6

8

a

b

c

a

b

c

a

ab

b

Mac

roph

yte

cove

r (%

)

0

20

40

60

80

100

a

b b

ba

b

a

b

a

Fig. 3 Biotic and abiotic ecological characteristics of the 13 reservoirs in relation to the occurrence of drying (black: did not dry out in 2009

or 2010; light grey: dried only in 2009; dark grey: dried in both 2009 and 2010). When ANOVA indicated significant differences between the

categories, each category differing by Tukey post hoc tests was given a different letter (a, b or c). Bars represent means (based on reservoir

means) + standard error. Broken vertical lines separate graphs of the three study years from that of 2004.



Tab

le2

ANOVA

resu

ltsfortheeffect

ofreservoir

categories

(reservoirsthat

did

notdry

outin

2009

norin

2010

;reservoirsthat

dried

outonly

in20

09;reservoirsthat

dried

outin

both

2009

and20

10)ontheab

iotican

dbioticecological

characteristicsofthestudiedreservoirs.

Theeffect

ofdryingwas

tested

forthebiomassoffish

,Microcystis,totalzo

oplankton

andDaphn

ia,percentagemacrophyte

cover,ch

lorophyll-a

(Chl-a)

concentration,totalphosp

horus(TP),totalnitrogen

(TN),su

spen

ded

matter(SM),Secch

idep

th,dissolved

oxygen

,

temperature,pH

andconductivity.Allen

vironmen

talvariablesweremeasu

redduringthesamplingcampaignsofSep

tember

2009

,20

10an

d20

11.In

addition,wealso

analysedthe

datacollectedin

Sep

tember

2004

byDejen

ieet

al.(2008).AllvariablesexceptpH

werelog-transform

ed

Param

eters

2009

2010

2011

2004

DF

SS

MS

FP-value

DF

SS

MS

FP-value

DF

SS

MS

FP-value

DF

SS

MS

FP-value

Fishbiomass(g;LOG)

237

.16

18.58

19.32

0.0003

220

.28

10.14

11.08

0.002

24.55

2.27

9.93

0.004

23.12

1.56

4.60

0.038

Microcystisbiomass

(lgC

L�1;LOG)

269

.79

34.89

8.51

0.006

237

.63

18.81

6.26

0.017

246

.14

23.07

7.03

0.012

20.23

0.11

1.48

0.270

Chlorophyll-a

(lgL�1;LOG)

21.90

0.95

9.079

0.005

21.00

0.50

10.72

0.003

20.92

0.46

5.74

0.021

20.38

0.19

4.79

0.058

Totalzo

oplankton

biomass

(lgL�1;LOG)

20.28

0.14

0.50

0.621

20.82

0.41

0.47

0.634

20.02

40.012

0.05

10.949

20.87

0.43

3.89

0.056

Dap

hnia

biomass

(lgL�1;LOG)

20.11

0.05

0.25

0.789

21.00

0.50

1.82

0.211

20.11

0.06

0.28

0.759

21.40

0.70

2.29

0.151

Macrophyte

cover

(%;LOG)

20.24

0.12

3.23

0.044

24.38

2.19

6.10

0.018

24.87

2.43

8.96

0.005

20.24

0.12

3.23

0.082

Secch

idisc

(m;LOG)

20.10

0.05

4.39

0.040

20.041

0.020

4.66

0.036

20.26

0.13

5.95

0.019

20.007

0.003

0.60

0.564

TPconcentration

(lgL�1;LOG)

20.99

0.50

5.50

0.024

20.66

0.33

5.95

0.019

20.87

0.43

6.25

0.017

20.44

0.22

3.55

0.068

TN

concentration

(lgL�1;LOG)

21.83

0.91

1.09

0.372

22.51

1.25

1.51

0.267

20.15

0.07

0.93

0.425

20.07

0.03

1.51

0.265

Susp

ended

matter

(lgL�1;LOG)

21.58

0.79

4.98

0.031

20.89

0.44

5.59

0.023

20.86

0.43

10.81

0.003

20.29

0.14

3.82

0.058

Dissolved

oxygen

(mgL�1;LOG)

20.012

0.006

2.447

0.136

20.04

0.02

12.12

0.002

20.01

70.008

4.96

0.031

20.01

0.008

1.05

0.383

Tem

perature

(°C;LOG)

20.005

0.002

2.22

0.159

20.006

0.003

1.736

0.225

20.00

10.0009

1.01

10.398

20.003

0.001

1.103

0.369

pH

20.302

0.151

0.362

0.704

20.248

0.124

0.506

0.617

20.13

80.069

0.29

80.748

22.344

1.172

6.441

0.015

Conductivity

(lSL�1;LOG)

20.044

0.022

0.889

0.441

20.037

0.018

0.535

0.601

20.01

70.008

1.12

40.362

20.137

0.068

2.547

0.127



dry in both 2009 and 2010 and the reservoirs that did not

dry at all (Fig. 3). Overall, zooplankton and Daphnia bio-

mass did not differ strongly between reservoir categories.

Longer-term impact of the propensity to dry: data from

2004

The propensity to dry was also found to be associated

with reservoir characteristics even in the absence of

recent drying. According to an RDA, the difference in

ecological variables measured in 2004 (Table S2) from 13

reservoirs that had not dried out for at least 2 years can

be explained by categories defined in 2009/2010

(Table 1). Ecological gradients between the drying and

permanent categories also show similarity with the pat-

terns observed for 2009–2011 (Fig. 2), although there are

some minor differences. For the period 2009–2011, the

pattern is very consistent, with reservoirs experiencing

drying having clear water, high oxygen concentration

and abundant macrophytes, whereas permanent

reservoirs had a high biomass of fish, Microcystis and

chlorophyll-a as well as high total phosphorus and sus-

pended matter. In 2004, the pattern was similar for most

variables, except for Microcystis biomass and oxygen,

which did not seem to be associated with probability of

drying. Furthermore, during 2004, total zooplankton bio-

mass was relatively high in permanent reservoirs.

The role of fish versus general effects of drying

We quantified the unique contribution of drying and

fish biomass on variation in environmental characteris-

tics using a variation partitioning following RDA

(Table 3; Fig. 4). The total combined variation explained

by both fish and drying was highly significant in all

study years, whereas the unique effects of drying and

fish were not significant in any of the three study years,

except 2004.

Discussion

Our results strongly suggest that the propensity to dry

out has important consequences for water quality of res-

ervoirs in the semi-arid regions of the cool tropics. This

has implications for both our understanding of reservoir

ecology in semi-arid regions as well as for reservoir

management.

Impact of drying on reservoir ecology

Complete drying affected key ecological characteristics

of the study reservoirs, resulting in lower nutrient

availability and phytoplankton, Microcystis and fish

biomass, and increased vegetation cover and water

transparency. Reservoir drying thus leads to an overall

increase in water quality. These observations are gener-

ally similar to those observed in temperate regions

(Scheffer, 1998; Van Geest et al., 2005; Jeppesen et al.,

2012). The observed differences in ecological characteris-

tics of the reservoirs were generally associated with dry-

ing and not with any other particular morphometric or

local characteristic of the reservoirs (such as altitude,

average depth or age). The reservoir categories differed

in surface area, with larger areas being more likely to

dry out (because they are more often used for irriga-

tion). The characteristics of the reservoirs that dried out

(low fish and phytoplankton biomass, higher transpar-

ency) are not typical for large compared with small res-

ervoirs, so it is safe to conclude that it is drying, not

surface area, which is the underlying cause of the

observed differences. Interestingly, two consecutive dry

years created a longer-lasting effect on macrophytes and

transparency than a single event. Whereas macrophyte

cover and water clarity resembled that in permanent res-

ervoirs within 1 year of the first drying event in those

reservoirs that dried only in 2009, the reservoirs that fell

dry twice still showed a greater coverage of macro-

Table 3 Result of variation partitioning following redundancy analysis testing for the relative importance of reservoir drying and fish

biomass on ecological characteristics of the 13 reservoirs studied in 2009, 2010 and 2011. Data from 2004 from Dejenie et al. (2008) are also

shown

Explanatory variable

2009 2010 2011 2004

DF Adj. R2 P DF Adj. R2 P DF Adj. R2 P DF Adj. R2 P

Fish effect 1 0.489 0.001** 1 0.193 0.031* 1 0.448 0.002** 1 0.049 0.563NS

Drying effect 2 0.451 0.007** 2 0.350 0.017* 2 0.452 0.006** 2 0.196 0.100NS

Fish + drying effect 3 0.468 0.010* 3 0.372 0.023* 3 0.507 0.009** 3 0.478 0.026*

Pure fish effect 1 0.017 0.238NS 1 0.021 0.269NS 1 0.054 0.141NS 1 0.282 0.019*

Common effect 0 0.471 NT 0 0.171 NT 0 0.393 NT 0 0 NT

Pure drying effect 2 0.020 0.543NS 2 0.179 0.059NS 2 0.059 0.191NS 2 0.528 0.018*

NT, non-testable; NS, non-significant, *significant at P < 0.05, **highly significant at P < 0.01.



phytes and had clearer water than permanent reservoirs

in 2011. Importantly, our data not only suggest that dry-

ing affects water quality in reservoirs, but show that the

propensity to drying has longer-term consequences for

the ecology of the reservoirs. This is most clearly seen in

the principal component analysis where, in all four sam-

pling campaigns, reservoirs belonging to the same cate-

gories cluster together (Fig. 2). This is intriguing because

the reservoir categories are identified based on the

behaviour of the reservoirs in both 2009 and 2010. While

it might be intuitive that the three categories of reser-

voirs show different ecological characteristics in 2010

and, if the effect of drying persists for more than 1 year,

in 2011, the fact that the same pattern was observed in

2009 and 2004 is more striking. Because all reservoirs

have been regularly visited since 2002, we know that

none of them fell dry in the 2 years preceding the sam-

pling campaign of 2004. Similarly, none of the reservoirs

fell dry in the period 2004–2008. The droughts of 2009

and 2010 were indeed exceptional for this region. In

2009, the reservoirs that fell dry only in 2009 differed

from those that also dried in 2010 in both fish and

Microcystis biomass (Fig. 3). This pattern is similar to

that observed in 2010. While the pattern is easy to

explain in 2010, the observations of 2009 pre-date the

second dry period (2010), suggesting that the reservoirs

differ also in the long term. For 2004, the ANOVAs indi-

cate that the major differences are among the permanent

reservoirs and the reservoirs that showed one or two

dry periods in 2009–2010. This is the case for fish and

zooplankton biomass, and also to some degree for the

amount of suspended matter. While statistically insignif-

icant, Fig. 3 also shows gradual changes in some envi-

ronmental variables (e.g. Microcystis biomass, TN, TP,

chlorophyll-a) between permanent reservoirs and those

that dried out once or twice. This also explains why the

multivariate analysis (Fig. 3) showed a clear gradient in

ecological characteristics with increasing propensity for

drying. The propensity to dry thus seems to affect the

ecology of the reservoirs over the longer term.

While, in general, the differences between reservoir

categories in 2004 were similar, even if reduced, to those

observed in 2009–2011, the patterns of Microcystis and

zooplankton/Daphnia biomass are noteworthy excep-

2009 Adjusted R-squared (Fish + Drying = 46.8% *)

Drying unique effect = 2.0%

Confoundedeffect = 47.1%

Fish unique effect =1.7%

Drying = 45.1% **Fish = 48.9% **

2010 Adjusted R-squared (Fish + Drying = 37.2%*)

Drying unique effect =17.9%

ConfoundedEffect = 17.1%

Fish unique effect = 2.1%

Drying = 35% *Fish = 19.3% *

2011Adjusted R-squared (Fish + Drying = 50.7% **)

Drying unique effect = 5.9%

Confounded effect = 39.3%

Fish unique effect = 5.4%

Drying = 45.2% **Fish = 44.8% **

2004Adjusted R-squared (Fish + Drying = 47.8% *)

Drying unique effect =52.8% *

0%Fish unique effect = 28.2% *

Drying = 19.6%Fish = 4.9%

Fig. 4 Result of variation partitioning following redundancy analysis testing for the relative importance of reservoir

dry-stand and fish biomass on ecological characteristics of the 13 study reservoirs in 2009–2011 and data of 2004. NT, non-testable; NS, non-

significant, * significant at P < 0.05, ** highly significant at P < 0.01.



tions. In 2009–2011, we observed that greater fish

biomass in the permanent reservoirs was associated with

higher chlorophyll-a and Microcystis but not with zoo-

plankton biomass. This suggests a bottom-up effect of

fish on phytoplankton (via nutrients) rather than top-

down control of zooplankton by fish. In 2004, a high fish

biomass in the permanent reservoirs was associated with

a high zooplankton biomass (Fig. 3) and was not related

to a higher chlorophyll-a and Microcystis biomass. This

pattern suggests that the zooplankton, that included

Daphnia, might have contributed to the control of phyto-

plankton and, particularly, Microcystis.

The effect of drying on reservoir characteristics may be

related to a drastic reduction in fish density (Scheffer,

1998; Beklioglu et al., 2007; Jeppesen et al., 2012). Alterna-

tively, it may be related to processes that are independent

of the presence of fish, such as a reduction in the carbon

pool by increased aerobic microbial activity (Corstanje,

2003), increased phosphorus uptake along with the oxida-

tion of iron (Scheffer, 1998; Moss, 2010), sediment com-

paction leading to reduced sediment resuspension (James

et al., 2001) and increased germination rates of macro-

phytes (Scheffer, 1998). The relation between drying and

fish biomass is unidirectional: drying is a key driver of

fish biomass, whereas fish biomass does not affect drying.

The global models in our variation partitioning analysis

explain a large proportion of the variation and are highly

significant (see Table 3; Fig. 4). The effects of fish and

drying are strongly confounded and cannot be separated,

as most of the variation explained by them is shared. This

indicates that reservoir variables, including fish, are

strongly affected by drying. In 2009–2011, neither the

effect of fish alone (corrected for effect of drying) nor the

effect of drying alone (corrected for the effect of fish)

explain a significant part of the variation in environmental

conditions. Thus, we cannot say that drying affects reser-

voirs independently of fish or that fish affect reservoirs

independently of drying. In the 2004 data set, both the

pure effects of fish and drying were significantly associ-

ated with the ecological characteristics of the reservoirs.

The 2004 data set thus provides evidence for separate

effects of fish and drying. These 2004 data suggest that

there is both an effect of fish independent of drying out as

well as a long-term effect of propensity to dry out that is

independent from changes in fish biomass.

Our results indicate that the propensity to dry out is

an important predictor of the ecological characteristics of

reservoirs in semi-arid regions. The impact of drying is

probably mediated by a major disturbance to the fish

assemblage as well as through an effect on nutrient

dynamics and sediment characteristics. Following dry-

ing, fish must recolonise the reservoirs and need time

the populations to increase. In our 2004 data set, there

were also pronounced differences in fish density

between permanent reservoirs and those that dry out.

The lower fish densities are probably a legacy of the dis-

turbance caused by occasional drying events.

Implications for reservoir management

Our results suggest that occasional drying should lead

to increased water quality, with fewer or less intensive

blooms, lower Microcystis biomass, lower biomass of

fish, higher transparency and more abundant aquatic

vegetation. Allowing reservoirs to dry out should there-

fore be considered as a management option for increas-

ing water quality and to prevent noxious cyanobacterial

blooms in eutrophic reservoirs such as these. This is

badly needed, as many reservoirs suffer from severe

(Dejenie et al., 2012) and toxic (Van Gremberghe et al.,

2011) Microcystis blooms, which may profoundly affect

the rural community, as the reservoirs are routinely

used for watering livestock. Microcystis can cause liver

diseases (Jochimsen et al., 1998; Codd et al., 1999).

Although not intended as a drinking water supply, the

reservoirs are occasionally also used by people for

drinking (Teferi et al., pers. obs.). Microcystine toxins

have also been observed to accumulate in fish meat (Ma-

galh~aes et al., 2003) and other components of the human

diet (Ibelings & Chorus, 2007) so that Microcystis blooms

may hamper exploitation of reservoirs for culturing fish.

Finally, microcystine toxins have even been observed to

produce toxic residues in crops irrigated with water con-

taining Microcystis blooms (Codd et al., 1999; McElhiney,

Lawton & Leifert, 2001; Chen et al., 2004; Peuthert,

Chakrabarti & Pflugmacher, 2007; Crush et al., 2008), so

even the use of reservoir water for irrigation poses a risk

unless water quality is improved. Our observation that

drawdown may help improve water quality is in line

with many observations, particularly in temperate sys-

tems (Jeppesen et al., 2012).

We could not distinguish between the separate effects

of fish removal and drying, so it is difficult to assess

whether fish removal per sewould already reduce the like-

lihood of Microcystis blooms. There is some circumstantial

evidence that fish removal improves water quality. First,

an enclosure experiment in two reservoirs suggested that

the presence of fish does promote Microcystis blooms

(Dejenie et al., 2009). Second, in the present study we did

observe a pure effect of fish presence on water quality for

the 2004 data set. However, it is likely that the effect will

be enhanced by drying, which is also evident from the



2004 data set. In addition, drawdown is the most efficient

way to remove fish from the reservoirs.

Our results provide some information on the fre-

quency with which drawdown should be implemented

to result in an improved water quality. First, we

observed long-term effects of drying in systems in which

we observed only one or two drying events in a period

of 8 years (2003–2011) and effects that persisted through

two wet years. This suggests that regular drying, as

observed here, in the order of perhaps one event every

5 years, might be sufficient to produce improvements in

water quality. On the other hand, we also found indica-

tions that two consecutive dry periods might result in

stronger and longer-lasting responses in terms of trans-

parency and macrophyte development.

With particular reference to the control of Microcystis

blooms, the data from 2004 do warn against too much

optimism. Although not significant, there was some

indication of higher Microcystis biomass in reservoirs

with a greater propensity to dry out. This could be

related to the greater densities of Daphnia in the more

permanent reservoirs, but how this is related to proba-

bility of drying, particularly given the higher fish bio-

mass in permanent reservoirs, is unclear. One possibility

is that phosphorus concentrations in the permanent res-

ervoirs are higher, thus supporting stronger develop-

ment of Daphnia, although this remains speculative.

Implementing drying as a strategy for improving

water quality may be an effective and relatively easy

management strategy for reservoirs in semi-arid regions.

However, there are several practical problems in a socio-

economic context. First, a forced drawdown may be

risky if the dry season is unexpectedly extended and

water is needed. However, a large quantity of water in

the reservoirs is currently lost because irrigation in

Tigray often starts only 4 months after the end of the

rainy season, from January till May/June (Eyasu, 2005).

Eyasu (2005) has calculated that this is not an efficient

use of the limited water stored in reservoirs. He suggests

starting irrigation soon after the end of the rainy season,

although the implications of such changes in irrigation

schemes for agricultural production still need to be evalu-

ated more carefully. We suggest that such a change in

irrigation might result in a win-win situation for both the

quality and quantity of water, with water being used

more effectively for irrigation and with higher water

quality in the subsequent season. However, the reservoirs

are also used for watering cattle and a forced drawdown,

combined with an extended dry season or a poor subse-

quent rainy season, may lead to problems for livestock

herders. Even though the reservoir water contains toxic

cyanobacteria that may be a health hazard to cattle, farm-

ers may still prefer this to having no access to water at

all. Therefore, the implementation of regular forced

drawdowns of reservoirs as a management strategy could

probably be implemented only if alternative sources of

water for cattle are available.

Acknowledgments

This study was financially supported by the Flemish

Institutional University Cooperation (IUC), under the

VLIR-UOS-Mekelle University IUC programme. We

thank the management team of the VLIR-UOS pro-

gramme for logistic support. We also thank Prof. Alan

Hildrew (Chief Editor) for his valuable help in the final

editing of the manuscript. This is publication no 5558 of

the Netherlands Institute of Ecology (NIOO-KNAW).

References

APHA (1999) Standard Methods for the Examination of Water

and Wastewater, 20th edn. American Public Health Associ-

ation, Washington DC.

Appelberg M. (2000) Swedish standard methods for sam-

pling freshwater fish with multi-mesh gillnets: stratified

random sampling with Nordic multi-mesh gillnets pro-

vide reliable whole-lake estimates of the relative abun-

dance and biomass of freshwater temperate lakes.

Fiskeriverket Information, 1, 3–32.

Beklioglu M., Altinayar G. & Tan C.O. (2006) Water level

control over submerged macrophyte development in five

shallow lakes of Mediterranean Turkey. Archiv f€ur Hydro-

biologie, 166, 535–556.

Beklioglu M., Romo S., Kagalou I., Quintana X. & Becares

E. (2007) State of the art in the functioning of shallow

Mediterranean lakes: workshop conclusions. Hydrobiolo-

gia, 196, 317–326.

Bell S.G. & Codd G.A. (1994) Cyanobacterial toxins and

human health. Reviews in Medical Microbiology, 5, 256–264.

Bottrell H.H., Duncan A., Gliwicz Z., Grygierek E., Herzig

A., Hillbricht-Ilkowska A. et al. (1976) A review of some

problems in zooplankton production studies. Norwegian

Journal of Zoology, 24, 419–456.

Bucak T., Saraoglu E., Levi E.E., Nihan Tavsanoglu €U., _Idil

C�akiroglu A., Jeppesen E. et al. (2012) The influence of

water level on macrophyte growth and trophic interac-

tions in eutrophic Mediterranean shallow lakes: a meso-

cosm experiment with and without fish. Freshwater

Biology, 57, 1631–1642.

Chen J., Song L., Dai J., Gan N. & Liu Z. (2004) Effects of

microcystins on the growth and the activity of superoxide

dismutase and peroxidase of rape (Brassica napus L.) and

rice (Oryza sativa L.). Toxicon, 43, 393–400.



Chiambeng G.Y. & Dumont H.J. (2005) The Branchiopoda

(Crustacea: Anomopoda, Ctenopoda and Cyclestherida)

of the rain forests of Cameroon, West Africa: low abun-

dances, few endemics and a boreal tropical disjunction.

Journal of Biogeography, 32, 1611–1620.

Chorus I. & Bertram J. (Eds) (1999) Toxic Cyanobacteria in

Water. A Guide to their Public Health Consequences. Monitor-

ing and Management. E & FN Spon, London.

Codd G.A., Metcalf J.S. & Beattie K.A. (1999) Retention of

Microcystis aeruginosa and microcystin by salad lettuce

(Lactuca sativa) after spray irrigation with water contain-

ing cyanobacteria. Toxicon, 37, 1181–1185.

Codd G.A., Ward C.J. & Bell S.G. (1997) Cyanobacterial tox-

ins: occurrence, modes of action, health effects and expo-

sure routes. Archives of Toxicology Supplement = Archive

f€ur Toxikologie Supplement, 19, 399.

Coops H., Beklioglu M. & Crisman T.L. (2003) The role of

water-level fluctuations in shallow lake ecosystems-

workshop conclusions. Hydrobiologia, 506, 23–27.

Corstanje R. (2003) Experimental and Multivariate Analysis of

Biogeochemical Indicators of Change in Wetland Ecosystem.

PhD Thesis, University of Florida, Gainesville.

Crush J.R., Briggs L.R., Sprosen J.M. & Nichols S.N. (2008)

Effect of irrigation with lake water containing microcys-

tins on microcystin content and growth of ryegrass,

clover, rape, and lettuce. Environmental Toxicology, 23,

246–252.

De Wit J. (2003) Stuwmeren in Tigray (Noord-Ethiopie): ken-

merke, sedimentatie en sedimentbronnen. MSc Thesis, Depart-

ment of Geography, KU Leuven.

Dejenie T., Asmelash T., De Meester L., Mulugeta A.,

Gebrekidan A., Risch S.R.S. et al. (2008) Limnological and

ecological characteristics of tropical highland reservoirs

in Tigray, Northern Ethiopia. Hydrobiologia, 610, 193–

209.

Dejenie T., Asmelash T., Rousseaux T., Gebregiorgis T.,

Gebrekidan A., Teferi M. et al. (2009) Impact of the fish

Garra on the ecology of reservoirs and the occurrence of

Microcystis blooms in semi-arid tropical highlands: an

experimental assessment using enclosures. Freshwater

Biology, 54, 1605–1615.

Dejenie T., Declerck S.A., Asmelash T., Risch S., Mergeay J.,

De Bie T. et al. (2012) Cladoceran community composition

in tropical semi-arid highland reservoirs in Tigray

(Northern Ethiopia): a metacommunity perspective

applied to young reservoirs. Limnologica, 42, 137–143.

Dumont H.J., Velde I. & Dumont S. (1975) The dry weight

estimate of biomass in a selection of Cladocera, Cope-

poda and Rotifera from the plankton, periphyton and

benthos of continental waters. Oecologia, 19, 75–97.

Eyasu Y.H. (2005) Development and Management of Irrigated

Lands in Tigray, Ethiopia. PhD Thesis, Wageningen Uni-

versity, Delft, the Netherlands.

Fl€ossner D. (2000) Die Haplopoda und Cladocera (ohne Bos-

minidae) Metteleuropas. Backhuys Publishers, Leiden.

Gasse F. (2000) Hydrological changes in the African tropics

since the last glacial maximum. Quaternary Science Review,

19, 189–211.

Getahun A. & Stiassny M.L.J. (1998) The fresh water biodi-

versity crisis: the case of Ethiopian fish fauna. SINET:

Ethiopian Journal of Science, 21, 207–230.

Graham J.L. (2007) Harmful Algal Blooms. US Department of

the Interior, US Geological Survey.

Haregeweyn N., Poesen J., Nyssen J., De Wit J., Mitiku H.,

Govers G. et al. (2006) Reservoirs in Tigray: characteristics

and sediment deposition problems. Land Degradation and

Development, 17, 211–230.

Havens K.E. (1993) Responses to experimental fish manipu-

lations in a shallow, hypereutrophic lake – the relative

importance of benthic nutrient recycling and trophic cas-

cade. Hydrobiologia, 254, 73–80.

Hillebrand H., Durselen L.D., Kirschtel D., Pollingher U. &

Zohary T. (1999) Biovolume calculation for pelagic and

benthic microalgae. Journal of Phycology, 35, 403–424.

Holdren C., Jones B. & Taggart J. (2001) Managing Lakes and

Reservoirs. North American Lake Management Society

and Terrene Institute, Madison, WI.

Horppila J., Peltonen H., Malinen T., Loukkanen E. &

Kairesalo T. (1998) Top-down or bottom-up effects by

fish: issues of concern in biomanipulation of lakes. Resto-

ration Ecology, 6, 20–28.

HTS (Hunting Technical Services) (1976) Tigray Rural Devel-

opment Study, Annex 2: Water Resources. Vol. 1: Hydrology

and Surface Water. Hunting Technical Services, Hemel

Hempstead (G.B.), 213.

Ibelings B.W. & Chorus I. (2007) Accumulation of cyanobac-

terial toxins in freshwater “seafood” and its consequences

for public health: a review. Environmental Pollution, 150,

177–192.

Iglesias C., Mazzeo N., Meerhoff M., Lacerot G., Clemente

J.M., Scasso F. et al. (2011) High predation is of key

importance for dominance of small-bodied zooplankton

in warm shallow lakes: evidence from lakes, fish exclo-

sures and surface sediments. Hydrobiologia, 667, 133–147.

James W.F., Barko J.W., Eakin H.L. & Helsel D.R. (2001)

Changes in sediment characteristics following drawdown

of Big Muskego Lake, Wisconsin. Archiv f€ur Hydrobiologie,

151, 459–474.

Jeppesen E., Sondergaard M., Jensen J.P., Havens K.E.,

Anneville O., Carvalho L. et al. (2005) Lake responses to

reduced nutrient loading- an analysis of contemporary

long-term data from 35 case studies. Freshwater Biology,

50, 1747–1771.

Jeppesen E., Søndergaard M., Lauridsen T.L., Davidson

T.A., Liu Z., Mazzeo N. et al. (2012) Biomanipulation as a

restoration tool to combat eutrophication: recent advances

and future challenges. Advances in Ecological Research, 47,

411–488.

Jochimsen E.M., Carmichael W.W., An J.S., Cardo D.M.,

Cookson S.T., Holmes C.E.M. et al. (1998) Liver failure



and death after exposure to microcystins at a hemodialy-

sis center in Brazil. The New England Journal of Medicine,

338, 873–878.

Komarek J. & Anagnostidis K. (2000) Cyanobacteria: Chloro-

coccales. Spektrum Academischer Verlag, Heidelberg-

Berlin, Germany.

Komarek J. & Anagnostidis K. (2005) Cyanoprokaryota:

Oscillatoriales. Spektrum Academischer Verlag, Heidel-

berg-Berlin, Germany.

Legendre P. & Legendre L. (2012) Numerical Ecology, Vol.

24, 3rd edn. Elsevier, Amsterdam.

Lemma G. (1996) Climate Classification of Ethiopia. Bulletin

3. National Meteorological Services Agency, Addis

Ababa.

Leps J. & Smilauer P. (2003) Multivariate Analysis of Ecologi-

cal Data Using CANOCO. University Press, Cambridge.

Lowe-McConnell R.H. (1966) Man-made Lakes. Academic

Press, London.

Magalh~aes V.F., Marinho M.M., Domingos P., Oliveira A.C.,

Costa S.M., Azevedo L.O. et al. (2003) Microcystins

(cyanobacteria hepatotoxins) bioaccumulation in fish and

crustaceans from Sepetiba Bay (Brasil, RJ). Toxicon, 42,

289–295.

McElhiney J., Lawton L.A. & Leifert C. (2001) Investigations

into the inhibitory effects of microcystins on plant

growth, and the toxicity of plant tissues following expo-

sure. Toxicon, 39, 1411–1420.

Meerhoff M., Clemente J.M., Teixeira-de Mello F., Iglesias

C., Pedersen A.R. & Jeppesen E. (2007) Can warm

climate-related structure of littoral predator assemblies

weaken the clear water state in shallow lakes? Global

Change Biology, 13, 1888–1897.

Mergeay J., Declerck S., Verschuren D. & De Meester L.

(2006) Daphnia community analysis in shallow Kenyan

lakes and ponds using dormant eggs in surface sedi-

ments. Freshwater Biology, 51, 399–411.

Moss B. (2010) Ecology of Fresh Waters: A View for the

Twenty-First Century, 4th edn. Wiley-Blackwell, London.

Naselli-Flores L. & Barone R. (2005) Water-level fluctuations

in Mediterranean reservoirs: setting a dewatering thresh-

old as a management tool to improve water quality. Hyd-

robiologia, 548, 85–99.

Nyssen J., Vandenreyken H., Poesen J., Moeyersons J.,

Deckers J., Mitiku H. et al. (2005) Rainfall erosivity and

variability in the Northern Ethiopian Highlands. Journal

of Hydrology, 311, 172–187.

Oksanen J. (2005) Vegan: R Functions for Vegetation Ecologists.

Available at: http://cc.oulu.fi/_jarioksa/softhelp/vegan.

html.€Ozen A., Karapınar B., Kucuk _I., Jeppesen E. & Beklioglu

M. (2010) Drought-induced changes in nutrient concen-

trations and retention in two shallow Mediterranean

lakes subjected to different degrees of management. Hyd-


Peres-Neto P.R., Legendre P., Dray S. & Borcard D. (2006)

Variation partitioning of species data matrices: estimation

and comparison of fractions. Ecology, 87, 2614–2625.

Peuthert A., Chakrabarti S. & Pflugmacher S. (2007) Uptake

of microcystins-LR and-LF (cyanobacterial toxins) in

seedlings of several important agricultural plant species

and the correlation with cellular damage (lipid peroxida-

tion). Environmental Toxicology, 22, 436–442.

SAERT (Sustainable Agriculture and Environmental Reha-

bilitation in Tigray) (1994) Sustainable Agricultural and

Environmental Rehabilitation Programme in Tigray, Vol. 1.

Prepared by UNDP (United Nations Development Pro-

gramme), UNECA (United Nations Economic Commis-

sion for Africa), Tigray Development Authority and the

Regional Government of Tigray, Mekelle, Ethiopia.

Scheffer M. (1998) Ecology of Shallow Lakes. Chapman and

Hall, London.

Scheffer M., Portielje R. & Zambrano L. (2003) Fish facilitate

wave resuspension of sediment. Limnology and Oceanogra-

phy, 48, 1920–1926.

Smirnov N.N. (1992) The Macrothricidae of the World. Guide

to the Identification of the Microinvertebrates of the Continen-

tal Waters of the World 1. SPB Academy Publishing, The

Hague, the Netherlands.

Smirnov N.N. (1996) Cladocera: The Chydorinae and Syciinae

(Chydoridae) of the World Guide to the Identification of the

Microinvertebrates of the Continental Waters of the World 11.

SPB Academy Publishing, The Hague, the Netherlands.

Søndergaard M., Jensen J.P. & Jeppesen E. (2003) Role of

sediment and internal loading of phosphorus in shallow

lakes. Hydrobiologia, 506, 135–145.

StatSoft, Inc. (2012) STATISTICA (data analysis software sys-

tem), Version 11. Available at: www.statsoft.com.

Teferi M., Declerck S.A.J., De Bie T., Gebrekidan A.,

Asmelash T., Dejenie T. et al. (2013) Ecology of the river-

ine Garra species (Teleostei, Cypriniformes) in reservoirs

of the semi-arid highlands of northern Ethiopia: tempo-

ral dynamics of feeding activity. Inland Waters, 3, 331–

340.

Teixeira-de Mello F., Meerhoff M., Pekcan-Hekim Z. &

Jeppesen E. (2009) Substantial differences in littoral fish

community structure and dynamics in subtropical and

temperate shallow lakes. Freshwater Biology, 54, 1202–

1215.

Tudorancea C. & Taylor W.D. (2002) Ethiopian Rift Valleys

Lakes. Backhuys Publishers, Canada.

Vallet-Coulomb C., Dagnachew L., Gasse F., Travi Y. &

Chernet T. (2001) Lake evaporation estimates in tropical

Africa from limited meteorological data. Journal of Hydrol-

ogy, 245, 1–18.

Van Geest G.J., Coops H., Roijackers R.M.M., Buijse A.D. &

Scheffer M. (2005) Succession of aquatic vegetation driven

by reduced water-level fluctuations in floodplain lakes.

Journal of Applied Ecology, 42, 251–260.



Van Gremberghe I., Van der Gucht K., Vanormelingen P.,

Asmelash T., Dejenie T., D’hond S. et al. (2011) Genetic

diversity of Microcystis blooms (Cyanobacteria) in

recently constructed reservoirs in Tigray (Northern Ethio-

pia) assessed by rDNA IT. Aquatic Ecology, 45, 289–306.

Van Wichelen J., Declerck S., Muylaert K., Hoste I., Geenens

V., Vandekerkhove J. et al. (2007) The importance of

drawdown and sediment removal for the restoration of

the eutrophied shallow lake Kraenepoel (Belgium). Hyd-


Warren A. & Khogali M. (1992) Desertification and Drought

in the Sudano-Sahelian Region 1985–1991. United Nations

Sudano-Sahelian Office (UNSO), New York.

Wetzel R.G. (2001) Lake and River Ecosystems, 3rd edn.

Academic Press, San Diego, CA, U.S.A.

Whitford L.A. & Schumacher G.J. (1973) A Manual of Fresh-

water Algae. Sparks Press, Raleigh, North Carolina.

Whitton B.A., Brook A.J. & John D.M. (2002) The Freshwater

Algal Flora of the British Isles: An Identification Guide to

Freshwater and Terrestrial Algae. Cambridge University

Press, Cambridge.

Zhou W., Pan X. & Kottelat K. (2005) Species of Garra and

Discogobio (Teleostei: Cyprinidae) in Yuanjiang (Upper Red

River) Drainage of Yunnan Province, China with descrip-

tion of a new species. Zoological Studies, 44, 445–453.

Supporting Information

Additional Supporting Information may be found in the

online version of this article:

Table S1. ANOVA results for the effect of reservoir cate-

gories (reservoirs that did not dry out in 2009 nor in

2010; reservoirs that dried out only in 2009; reservoirs

that dried out in both 2009 and 2010) on the morpho-

metric and regional characteristics of the studied reser-

voirs.

Table S2. Results of the redundancy analysis: amount of

variation in ecological variables explained by reservoir

drying for the sampling campaigns of 2009, 2010 and

2011, and for the data collected in 2004 by Dejenie et al.

(2008).

Figure S1. Map of Ethiopia and the location of the 13

reservoirs in Tigray region, northern Ethiopia.

Figure S2. Morphometric and regional characteristics of

the 13 reservoirs in relation to drying (black: no drying

in 2009 nor 2010; light grey: drying only in 2009; dark

grey: drying in both 2009 and 2010).

(Manuscript accepted 9 December 2013)



Tefferi 2014_Drying Effects Tropical Reservoirs

Documents