International Journal of AgriScience Vol. 2(9):810-830, September 2012 www.inacj.com ISSN: 2228-6322© International Academic Journals

Opportunistic vascular introductions in agricultural wetlands of East Africa

Handa C.1, Alvarez M.2, Becker M.3, Oyieke H.1, Möseler B.M.2, Mogha N.4, Kamiri H.5 ¹National Museums of Kenya, Nairobi, Kenya; *Author of correspondence (email: [email protected]) ² Vegetation Ecology, INRES, University of Bonn, Germany;3 Crop Science, INRES, University of Bonn, Germany 4Botany, DUCE, University of Dar es Salaam, Tanzania;5Karatina University College, Kenya

Received July 2012; accepted in revised form August 2012

ABSTRACT The impact of wetland cultivation was investigated on species introductions using species functional traits. The roles of hydric regimes were assessed on two floodplains and two inland valleys in Kenya and Tanzania. Vegetation was assessed in 224 plots, each measuring 100 m2. Land use gradients were categorised as unused, grazed, fallow and cultivated. Species presence and cover were recorded in each plot. Data on species functional traits including Raunkiaer’s life form were collected both in the field and from related literature. Indicator species analysis (ISPA) was done to classify weeds relative to land use gradients. Canonical correspondence analysis (CCA) was done on soil environmental variables in relation to weeds. Of the species recorded, 78% were classified as weeds. Fallow and cropped plots had higher weed species richness compared to both unused and grazed plots (p < 0.017). Therophytes constituted 70% of total weed presences. Five indicator weed species were identified: Cynodon dactylon ((L.) Pers)) on completely drained potassium rich plots; Leersia hexandra (Sw) on wet to moist with high organic C and N; Malva parviflora (L) on cultivated inland valleys; Oxalis corniculata (L) on fallow and cultivated inland phosphorous rich maize plots within inland valleys and Typha capensis ((Rohrb.) N.E.Br.) in saline and alkaline plots. Raunkiaer’s life forms, life history and growth forms are good indicators of disturbance in wetlands. Results demonstrated that while wetland disturbance from agricultural activity eliminates native species, a regime shift acts in favour of introduced weed species.

Keywords: obligate uplands, obligate wetlands, Raunkiaer’s life form, species additions, weeds

INTRODUCTION eastern Africa, and now poses a real threat Population growth, infertile upland soil and to their continued existence (Dixon and erratic climatic conditions are some of the Wood 2003; Harper et al. 1990). reasons behind the increasing use of Disturbances to wetlands, especially hydro- wetlands in eastern Africa for agriculture period alteration (Harper et al. 1990) from (Kyle and Leishman 2009; Schuyt 2005; agricultural development and human Wood and van Halsema 2008). settlement, create opportunities that are well Consequently, many small wetland areas matched by opportunistic wetland plant have been drained for subsistence and species (Howard and Chege 2007; Zedler commercial agriculture; an action that has and Kercher 2004). This is the situation in overstretched small wetland resources in small wetlands currently used for agriculture

International Journal of AgriScience Vol. 2(9):810-830, September 2012 810

in eastern Africa; they become devoid of modification such as wetland drainage can native plant communities, even though they only lead to elimination, reduction or local can still perform limited ecological extinction of species, thus promoting functions. The majority of these wetland invasibility. On a global scale for example, areas are characterized by the presence of there is a direct relationship between species introduced weeds such as sedges and grasses decline and anthropogenic related stress on in old fallow or abandoned plots and natural ecosystems (Perrings et al. 2010). herbaceous in cultivated areas (Handa Little research has focused on species 2011). These introduced plants that invade additions due to landscape modification and natural areas such as wetlands sometimes ecological regime shifts arising from threaten local native communities and anthropogenic activities such as wetland reduce the diversity of native species agriculture in eastern Africa. Species gains altering an ecosystem’s function (Daehler have been observed especially at regional 1998). The ecological significance of and local levels as range-expanding habitat species’ diversity within an wetland generalists invade these disturbed ecosystem used for agriculture has recently environments (Hobbs and Mooney 1998). become a popular topic for related empirical Such weeds invade new species’ pools, studies (Pollnac et al. 2009). The current typically at the expense of rare and often consensus among ecologists and endemic, native species that disappear conservationists in eastern Africa is that (Hobbs and Mooney 1998; Rooney et al. degradation of wetlands from agricultural 2007). This element of species additions in a activity is promoting a loss of biodiversity modified landscape could at times be greater (Dixon and Wood 2003; Hall et al. 2009; than losses, which might result in significant Harper et al. 1990). Even though tropical effects on ecosystems and some of the biotic sedge dominated wetlands are known to be components that they invade (Hobbs and species poor in flora (Chapman et al. 2001), Mooney 1998). In oceanic islands for most wetland vegetation in East Africa plays example, the number of naturalizations has an important role as a habitat for fauna such increased dramatically for vascular plants, as migratory water birds (Maclean et al. with numbers of naturalizations greatly 2006). However, the fact that permanent exceeding those of extinctions (Sax et al. wetlands are species poor makes them 2007). The risk is that the impact of some of vulnerable to new species additions these introduced species (e.g. invasive especially when a hydrological regime is species) may be irreversible if not detected altered (Van der Valk 1981; Zedler and well in advance (Doren et al. 2009; Gordon Kercher 2004). The introduction of new 1998). species is what we call invisibility, the term Despite the well known impact of non- describes the susceptibility of an native plants on ecosystems (Gordon 1998), environment and in this particular case a their adaptive capabilities, especially in wetland environment to the colonization and transition ecosystems such as wetlands, have establishment of individuals from species not yet been well researched and remain not currently part of the resident community little understood. Over time, these exotic (Davis et al. 2005). Traditionally, more species may alter a wetland ecosystem diverse communities are less susceptible to because of their ability to self rejuvenate and invasion than species-poor communities establish resilience. Understanding trends in (Lodge 1993). However, there is a the spread and density of weeds and exotic widespread assumption that landscape species, including the impact of control and

811 International Journal of AgriScience Vol. 2(9):810-830, September 2012 management activity, is therefore necessary Gilbert-Carter 1937) to study the impact of to manage affective restorative strategies of biophysical changes on species affected ecosystems (Hulme 2006; Zedler establishment. For example, soil disturbed and Kercher 2004). sites tend to have more weed vegetation that Studying composition of weed species alone are predominantly therophytes (Lososová et may not be adequate to understanding those al. 2004) and wind dispersed species ecosystems experiencing frequent compared to less disturbed sites that are perturbations (Richardson et al. 2007). One often rich in geophytes, chamaephytes, of the simplest yet underutilized tools is the phanerophytes with greater percentages of use of plant functional traits (PFTs). PFTs vegetatively reproducing species (McIntyre are morphological, physiological or et al. 1995). phenological traits that reflect plant Hypothetically, weed vegetation that performance. The study of PFTs may assist predominantly consists of annual plants in making predictions relating to the shows a much higher degree of temporal dynamics of ecosystems such as wetlands dynamics than other vegetation types, such (Violle et al. 2007). PFTs are important as those that operate both on the scale of ecological tools as they provide an seasonal changes (Lososová et al. 2004) and indication of a plant community’s response long-term and this corresponds to gradually to variations in ecosystem attributes and increasing intensification of agricultural processes that is largely independent of a production (Andreasen et al. 1996). Thus taxonomic bias (Kyle and Leishman 2009). agricultural activity in wetlands leads to loss PFTs for example have been used in the past of species and can also lead to establishment as a tool for the restoration and of additional weed species. Thus, weed rehabilitation of ecosystems such as functional traits can be used as indicators of degraded wetlands (Kyle and Leishman regime shift from a natural wetland to a 2009). Another advantage of the using PFTs disturbed wetland used for agriculture. This is that they have also been extensively used paper therefore seeks to assess the impact of to predict the invasiveness of species and the wetland regime change on introductions of invasibility of habitats in the past (Pyšek and weed species using species functional traits Richardson 2007; Richardson et al. 2007) and changing hydrological conditions. and can thus be replicated in a wide MATERIALS AND METHODS geographical range. They have also been used to gauge ecological processes and to Study sites examine impacts of fire and grazing on plant This study was done on two representative communities (McIntyre et al. 1995). For floodplains and two inland valleys (Fig. 1). instance there is a relationship between soil The floodplains comprised one highland disturbed sites and spectrums of life- floodplain in Kenya (Ewaso Narok swamp) histories (McIntyre et al. 1995). The most and one lowland floodplain in Tanzania common PFT used to classify species in (Mkundi-Sangei swamp along the River relation to human disturbance has been Mkomazi). Highland inland valleys used in growth or life form, as this trait tends to the study were the Tegu wetland on the foot correlate with other physiological and slopes of Mt. Kenya and the Lukozi valley morphological traits. Most of these life form in West Usambara mountains. While Tegu traits respond differently to disturbance and Lukozi were typical inland valley levels. In this paper we have used wetlands, Ewasonarok and Malinda were Raunkiaer’s life form traits (Raunkiaer and riverine-floodplains, with origins from the

International Journal of AgriScience Vol. 2(9):810-830, September 2012 812

Fig 1. A map (not drawn to scale) showing study sites (in boxes); Tegu wetland (Mt. Kenya), Ewaso Narok (Aberdares), Mavumo-Hemboye-Hambalawey (West Usambara) and Malinda (Pangani plains). Inset is the map of Africa with eastern Africa appearing as dark shaded. Modified from Handa (2011).

Vegetation sampling rivers Ewasonarok and Mkomazi A list of all species introduced into the wetlands was made from an assessment of respectively. All these wetlands were under 2 crop cultivation and livestock farming. The 224 plots each measuring 100 m in two remaining undisturbed parts of wetlands floodplains and two inland valley swamps in were largely dominated by papyrus Kenya and Tanzania. Four land use (L.) in both highland and lowland flood gradients were identified, namely plains in Kenya and Tanzania. The inland undisturbed plots as reference sites under valleys however, were characterised by either Cyperus papyrus (L) or Typha diverse species of e.g. capensis ((Rohrb.) N.E.Br.) strands, grazed, Schoenoplectus corymbosus (Roth ex Roem. fallow and cropped (e.g. maize, cauliflower, & Schult.) and Cyperus dives (Delile). The tomatoes, and arrowroots). In each plot, land that was left fallow was dominated by species presence and percentage cover were upland weeds and permanently flooded recorded for herbs, sedges, grasses, vines, sections with weeds such as Azolla pinnata shrub and trees. Un-identifiable species were collected and placed in a plant press for (R.Br). In the floodplains, there was also intensive pastoral livestock activity as in the further identification either at the National case of Ewaso Narok wetland and River Museums of Kenya (NMK)’s East African Mkomazi swamps in the Pangani basin. Herbarium (EA) or at the University of Dar- es salaam’s herbarium by expert botanists.

International Journal of AgriScience Vol. 2(9):810-830, September 2012 813

Data on functional traits of all species laboratory analysis (Okalebo et al. 2002). including Raunkiaer’s (Raunkiaer and Laboratory analysis was done on soil Gilbert-Carter 1937) life form, habit, samples at the labs of Plant Nutrition, duration and taxonomical group were Department of the Institute of Crop Science collected. Raunkiaer’s life form subdivisions and Resource Conservation of the (Table 1) are based on the location of a University of Bonn (Kamiri 2010) using plant’s growth point (bud) during seasons standard methods recommended for tropical with adverse conditions (cold seasons, dry soil samples. Total soil C and N were seasons), first proposed by Raunkiaer in determined by an automated CNS elemental 1934 (Raunkiaer and Gilbert-Carter 1937). analyser (Euro EA 3000, manufacturer: Euro Based on floristic attributes, all weed Vector SpA, Milan, Italy). Available species including potentially invasive phosphorus and exchangeable potassium species were recorded and described. were analysed by Modified Olsen P method Information on PFTs and species affinity to (Olsen et al. 1954) using a 1:10 soil/solution ratio. The amount of potassium in the extract wetland was obtained from field collections and observations. Other sources from was measured on the Eppendorf Elex 6361 literature used for this research are as flame photometry and phosphorus was follows (Agnew and Agnew 1994; Haines measured colorimetrically (molybdenum and Lye 1983; Turrill and Beentje 1952- blue) by an Eppendorf ECOM 6122 2007) and East African herbarium spectrophotometer. The volumetric soil moisture level was determined in every plot collections at NMK, Nairobi. Online databases such as Aluka-Jstor (ALUKA at the time of vegetation assessment 2010) , PROTA 4U (Protabase 2010) and (beginning of rainy season rainy season) using Time Domain Reflectometry (TDR) flora of Zimbabwe (Hyde and Wursten 2 2011). The indicator status (i.e. affinity to with 10 readings collected per plot 100 m wetland) of each weed species was plot and later averaged. Soil salinity was determined for obligate and facultative represented as electrical conductivity (EC) wetland species (Tiner 1999). Indicator and soil pH levels were determined in 1:2.5 status is a species probability of occurrence soil water extract using EC and pH meters at in a wetland as opposed to an upland habitat; room temperature (22° C). for example Obligate wetlands (OBL) are Data analysis species having a percentage of probability of Weed diversity occurrence in a wetland of > 99%; Introductions of new species were facultative wetland species (FACW), 67% - determined in terms of numbers of plant 99%; facultative species (FAC), 34-66%; species not naturally occurring in wetlands facultative upland species, 1- 33%; and of East Africa (Ervin et al. 2006b) and obligate upland species (UP), < 1% (Lopez expressed as percentages. Comparisons of and Fennessy 2002; Mack 2007; Tiner weed functional traits were made between 1999). wetland and non-wetland species, based on Soil sampling and analysis species-level traits, presence and percentage In every subunit, a composite sample of 9 cover. Weed functional traits including top soil samples (2-15 cm) were collected Raunkiaer’s life form system, growth form (Mack 2007). Sampling was done in a ‘W’ and lifespan within each wetland and land shape using a 5 cm diameter-hand-held use were compared using a non-parametric auger. The soil was air-dried, ground and statistical analysis and PASW statistics sieved to pass a 2 mm sieve for subsequent version 18.1 where homogeneity of the

International Journal of AgriScience Vol. 2(9):810-830, September 2012 814

Table 1. Plant functional traits used to classify weed species including Raunkiaer’s life forms Traits Attributes Growth form herb, sedge, grass, shrub, tree, and duration Lifespan annual, perennial Raunkiaer’s life form Therophytes Annual (monocarpic) plants, includes some facultatively perennial plants (polycarpic) that were judged to be predominantly annual. They persist through seeds Geophytes persistent buds buried to a depth of 2-3 cm Chamaephytes persistent buds > =1 cm and < 20-30 cm above ground surface Phanerophytes persistent buds >20-30 cm on stems above in the ground, includes twiners and vines Hemicryptophytes persistent buds are in the immediate vicinity of the soil surface only, maximum height, 1cm Helophytes not free plant: growing in water but attached to bottom

mean was not met. The rate of occurrence or which a species is distinguished between relative frequency (f) of all weed species groups in terms of relative frequency and recorded in the four wetlands were evaluated relative abundance (Flinn et al. 2008; according to the formula: f=ss/ts, where ss is McCune and Grace 2002; McCune and the number of sites in which the species Mefford 2006). The significance of each occurred and ts the total number of sampling indicator value was tested using units (Goettsch and Hernández 2006). multiresponse permutation procedures Indicator Species Analysis (ISPA) (MRPP) based on the Bray-Curtis index To determine indicator species, firstly, with 4,999 permutations at the random seed groupings of 330 weed species were number of 5,768. Group definition, identified using a hierarchical agglomerative differences were determined using cluster analysis using Bray-Curtis similarity multiresponse permutation procedures indexes based on percentage covers. Only (MRPP) based on Sørensen (Bray-Curtis) those species were recorded that had similarities using PCORD 5 (McCune and occurrence in more than three plots. A Grace 2002; McCune and Mefford 2006). flexible beta linkage method with Bray Vegetation types were named according to Curtis as the distance measure, with β = the weed species with the highest indicator −0.25 was employed in the program PC- value. All analyses were done using PCORD ORD (McCune and Grace 2002). The group version 5 (McCune and Grace 2002; membership variable was set at five, given McCune and Mefford 2006) and PASW that there were four wetlands and four land version 18 (SPSS 2010). use types used in the study. Indicator species Ordination: The spatial relationship analysis (Dufrene and Legendre 1997) was between soil variables and introduced then used to choose an appropriate number species was tested using Canonical of groups from results of the cluster Correspondence Analysis (CCA) (McCune analysis. ISPA calculates indicator values and Grace 2002). Axis scores were for each species in each group as the product standardized by Hill’s method (Hill 1979). of a species’ mean abundance in that group All datasets were revitalised prior to relative to other groups and the proportion statistical analysis (McCune and Grace of sites in that group where it is present. The 2002). indicator values thus represent the degree to

International Journal of AgriScience Vol. 2(9):810-830, September 2012 815

RESULTS Results of the cluster analysis yielded five distinct cluster groups for indicator weed Weeds diversity species in the four wetlands, with 15% of Out of the 425 species that were sampled, information retained in the dendrogram. 78% (n=330) were classified as weed There were significant differences (MRPP) species representing 66 families. Highland among the five indicator groups with an floodplains had 159 weed species, whereas overall T= -66 and within group agreement lowland floodplains of Malinda registered A=0.44, p = 0.000. From the indicator 154 weed species. The highland inland species analysis, five groupings were created valley in Tegu registered 150 weeds while as follows; Cynodon dactylon ((L.) Pers)) the mid-montane inland valley of Lukozi (group code 1), Leersia hexandra (Sw) registered 133 species. There were no (group code 31), Malva parviflora (L), significant differences between floodplains (group code 54); Oxalis corniculata (L) with respect to introduced species (group code 56) and Typha capensis composition (p > 0.05). However, the two ((Rohrb.) N.E.Br.) (group code 85). flood plains were significantly different Cynodon dactylon ((L.) Pers)) indicator from the mid-montane inland valley (p< weed group was composed of 58 plots. The 0.05). Of the 330 weeds, 70% were dicots group registered 46 species associated to it. while the remaining 30% were monocots. Some species associated with this group had The average weed species richness was 18 ± high indicator values; Schkuhria pinnata 2 9 species per 100 m plots. There were ((Lam.) Thell.), Oxygonum sinuatum dissimilarities in species composition among ((Hochst. & Steud. Dammer), and Tagetes wetlands in inland valleys and floodplains; minuta (L). This group was associated with so there were differences along use high phosphorous and potassium contents gradients. Fallow and cropped plots had but with lower moisture content levels in the significantly higher weed species soil. Land uses associated with this group composition compared to either grazed or were mainly intensively cultivated (cropped) unused portions (p = 0.0177). Average weed patches, fallow plots in highland floodplains diversity in all land uses had a similar (Ewaso Narok) and a few plots under crops pattern to fallow and cropped areas from inland valleys (Fig. 3). There was no recording more weed flora compared to contribution from lowland floodplains. either grazed or unused areas of wetland Leersia hexandra (Sw) indicator weed (Fig. 2). Over 37% (n = 121) of all weed group was composed of 59 weeds in 53 species recorded had relative frequency of plots, mainly from lowland flood plains of occurrence below 1% i.e. many appeared in Malinda, Sangei and Mafuleta. A few plots less than three plots sampled. from uncultivated portions in the highland Examples of common weeds encountered inland valleys and highland floodplains were are shown in Table 2. also skewed to this group and positively and Potentially invasive species registered on negatively to axis 2. This group was closely fallow and cultivated plots were Vossia associated with high soil moisture content, cuspidata (Griff.), Pistia stratiotes (L) and, high total carbon and nitrate levels. Some Azolla pinnata (R.Br). Mimosa pigra (L) group members included Azolla pinnata was common in fallow and seasonally (R.Br), Ageratum conyzoides(L), Ethulia flooded areas whereas Datura stramonium conyzoides (L.f. ), Boerhavia diffusa (L.), (L) was common in completely drained and Basilicum polystachyon (L.), and Rotala cultivated plots. fluitans (Pohnert). Indicator species analysis

International Journal of AgriScience Vol. 2(9):810-830, September 2012 816

Fig 2. Average numbers of weed species in relation to land use type in four wetlands. FP=Flood plain; IV= Inland valley.

Table 2. A list of cosmopolitan weed species in the four wetland areas. Species occurrence and frequency is shown. f = frequency.

Weeds Occurrence f Invasibility gradients Ageratum conyzoides (L) 88 39 Weed Crops Amaranthus hybridus (L) 115 51 Weed Fallow Bidens pilosa (L) 116 52 Invasive Cropped Commelina benghalensis (L) 168 75 Invasive cropped Cynodon dactylon ((L.) Pers)) 119 53 Weed Grazed Cyperus rotundus (L.) Palla 136 60 Weed cropped Galinsoga parviflora (Cav) 130 58 Weed Cropped Harpagocarpus snowdenii (Hutch. & Dandy) 67 30 Weed Cropped Leersia hexandra (Sw) 105 47 Weed Fallow Portulaca oleracea(L) 100 44 Weed Cropped Spilanthes mauritiana (A.Rich.) DC. 69 31 Weed Fallow Vernonia poskeana ( Vatke & Hildebr.) 69 31 Weed Fallow Typha capensis((Rohrb.) N.E.Br.) 34 Invasive Fallow

Malva parviflora (L) indicator weed group officinale (Weber), Setaria verticillata ((L.) was composed of 45 sampling units with 44 P. Beauv.)), Cyperus rotundus (L.) Palla)) species, all from mid-montane inland and Eleusine indica ((L.) Gaertn.). This valleys-the only distinct wetland group group was positively skewed to axes 1 and 2 without a contribution from other wetlands. and strongly associated with pH and Other group members with high indicator electrical conductivity. values included Chenopodium album (L.), Oxalis corniculata (L.) indicator weed Chenopodium opulifolium (Schrad.), group had 39 plots all from the highland Chenopodium murale (L.), Taraxacum

International Journal of AgriScience Vol. 2(9):810-830, September 2012 817 inland valley of Tegu except two fallow plots from

Fig 3. Canonical correspondence analysis (CCA) biplot for weed species (n = 330) in four wetland types (n=201) categorized into five groupings from indicator species analysis (1= Cynodon dactylon ((L.) Pers)) group, 31= Leersia hexandra (Sw) group, 54 = Malva parviflora (L) group, 56 = Oxalis corniculata (L.) group and 85 = Typha capensis group). Soil environmental variables are represented by vectors. EC = Electrical conductivity; MO = percent moisture; C = Total carbon and P = available phosphorous and K = exchangeable potassium. The two axes together explained 34% of the variability among the five groups from analysis of indicator species. the mid montane inland valley that was was from the mid-inland valley. The group associated with this group. Other group was associated with high phosphorous levels members included Commelina benghalensis in soils but negatively skewed to axes 1 and (L), Centella asiatica (L) Urb)), Galinsoga 2. urticifolia ((Kunth) Benth.), Vernonia Typha capensis ((Rohrb.) N.E.Br.) indicator poskeana (Vatke & Hildebr.), Kyllinga alata weed group was composed of 6 sampling (Nees), and Stellaria media (L.Vill.). This units all from mid-montane inland valleys group was dominated by highland inland and lowland floodplains and characterised valley sampling units; mainly from cropped by high electrical conductivity and areas and fallow plots. Other contribution

International Journal of AgriScience Vol. 2(9):810-830, September 2012 818

permanently inundated. Three other species Facultative upland and obligate upland were were associated with this group, namely the most dominant weed categories relative Celosia trigyna (L.), Epilobium hirsutum to flooding regimes (Fig. 6 and 7). Fallow (L.) and Paspalum vaginatum (Sw.). This and cultivated plots had greater percentages group was positively related to axes 1 and 2 of species under either UP or FAC. Obligate of the biplot. wetland (OBL) and facultative wetland (FACW) categories however were Canonical Correspondence Analysis associated with unused plots. Most plants A 2-D CCA biplot shows the distribution of species (77%) in the four wetlands surveyed sampling units relative to species were either facultative wetland (FAC) or occurrences, abundance in various land use upland (UP) weeds. Lowland floodplains types in relation to soil parameters (Fig. 3). had more species with high affinity to Lifespan permanent inundation (i.e. FACW and OBL) With the exception of cropped subunits, compared to inland valleys, which had a there were significant differences between bigger proportion of upland weed species. perennial and annual weeds across the Raunkiaer’s life forms and land use wetlands relative to land use (p < 0.05). There were significant differences in Cultivated and fallow plots within the Raunkier’s life form traits between land use wetlands had higher occurrences of annual types (p=0.05). Weed species in cultivated weeds compared to either grazed or unused lands were mainly therophytes with a higher plots (Fig. 4). affinity to completely drained areas. Growth forms Helophytes and hydrophytes however were The relationship between land use and common in unused plots and mainly growth form shows that herbs were the most constituted perennial weed species in either abundant weeds across the wetlands with permanently or seasonally inundated higher frequencies (Fig. 5). Fallow and wetland areas. They were mainly associated cropped plots had significantly higher with rice plots in lowland flood plains. frequency of occurrences of herbs and Individual wetlands showed no significant grasses compared to either unused or grazed differences relative to life form traits (p< plots. 0.05). However, mid-montane inland valleys With respect to growth forms, a greater had a lower proportion of helophytes as percentage of species additions were herbs weeds compared to the other three wetland and grasses, the majority of which were areas (Fig. 8). weeds (Fig. 5). Only 5% of herbs were DISCUSSION classified as non weeds. Of all the herbs East Africa has experienced an unusually recorded, 82% were weed species, 13% of fast transformation of small wetlands from the herbs recorded were identified as extensive subsistence cropping to intensive potentially invasive. Sedges had no invasive vegetable production (Wood and van species but constituted 57% of the species Halsema 2008). Under this transformation recorded as weeds. A greater percentage of crops have replaced emergent aquatic the grasses within the wetland were upland macrophytes in more than 70% of the grasses therefore qualified as weeds. There wetlands (Dixon and Wood 2003; Schuyt was a perfect negative correlation between 2005). Because they are drained, when left weeds and non-weeds (p < 0.027) but a fallow these wetlands attract obligate upland higher positive correlation between weed (UP) species. One of the visible aspects of richness and herbaceous richness.

819 International Journal of AgriScience Vol. 2(9):810-830, September 2012

Fig 4. Frequency of annual and perennial weed species in relation to land use types in four wetlands. FP= flood plain; IV= inland valley this altered wetland structure is the addition richness of new additions recorded in used of species to the flora. This has been wetlands is indicative of a regime shift from documented in the uplands where a change permanently inundated, sedge dominated from shifting to intensive/permanent wetland to seasonally inundated annual agriculture in the tropics eliminates many herb-dominated wetland (Hobbs and woody, secondary species subsequently Mooney 1998; Rooney et al. 2007). Related replacing them with aggressive, herbaceous to new additions is the rich plant growth that pantropical elements (Kellman 1980). These is not indicative of high fertility of wetland species additions in most cases are areas but of quick nutrient recycling and agricultural weeds whereas others are high nutrient use efficiency (Junk 2002). considered environmental weeds by virtue of misplaced occurrences; in these cases in Wetland types cultivated small wetlands. Despite the fact that these results show no significant differences (p > 0.05) on species

traits between wetland type, floodplains Weed species additions These results show that there has been were richer in introduced species in immense species additions in the four comparison to the inland valley wetlands. wetlands selected for this study, which is in The replacement of native habitats by agricultural fields does indeed facilitate the agreement with other research that has also concluded that agriculture in wetlands does easy movement of propagules (Tickner et al. indeed lead to an ecological regime shift 2001). Thus, species that are readily (Folke et al. 2004). The high weed species dispersed by water and whose seeds,

International Journal of AgriScience Vol. 2(9):810-830, September 2012 820

Fig 5. Frequency of weeds growth forms in relation to land use types in four wetlands: a) Highland and lowland floodplains (FP); b) Highland and mid-montane inland valleys (IV)

diaspores or whole plants establish well on Valk 1981). The fact that inland valleys are bare ground, are best equipped to invade perennially wet/or moist; cushions them bare spaces along stream banks and against additional obligate upland floodplains (Stromberg et al. 2007; Van der introductions compared to floodplains,

821 International Journal of AgriScience Vol. 2(9):810-830, September 2012

Fig 6. Frequency of growth forms in relation to weeds and non-weeds (n=425). The frequencies are represented by percentages of the total number species in each growth form and not a percentage of total number of species sampled (n=425).

whose seasonal flooding in combination soil, nutrients and water are often lost at with manmade drainage channels is rates faster than they are replaced. conducive to establishment of these ISPA and weeds opportunistic species. Differences in weed diversity observed The five indicator species groups that were across wetlands in relation to land use type generated were grouped on the basis of can also be attributed to different nutrient availability, moisture content or management practices that were specific and seasonality of flooding. The weeds were unique to every wetland type assessed in the somehow biased towards those land uses study (Pyšek et al. 2005; Pyšek and based on existing cropping systems. Richardson 2007). This high diversity of Cynodon dactylon ((L.) Pers)) species- weed species in used wetlands however does group was a representation of weeds not indicate a healthy wetland but rather a accustomed to dry habitats with less reduced wetland quality due to an increased moisture content, such as facultative upland disturbance. The weed communities on (FC) species and obligate upland species. wetlands contain neither the biodiversity nor Other characteristics of this group included the aggregate adaptive ability to coalesce an affinity to maize and beans fields. into self-sustaining, self-replicating systems Tagetes minuta (L), Oxygonum sinuatum hence they have low resilience (Patchett and ((Hochst. & Steud. Dammer)), Bidens pilosa Wilhelm 1999). This low resilience (L) and Schkuhria pinnata ((Lam.) Thell.) therefore indicates that the destiny of many were especially common species of this systems dominated by weeds is further group. The common Raunkiaer’s functional destabilized, during which resources such as trait of this group was therophytes, the

International Journal of AgriScience Vol. 2(9):810-830, September 2012 822

Fig 7. Frequency of weeds and non weeds in relation to their affinity to flooding regimes. FAC=facultative; FACU=Facultative upland, UP= obligate upland; FACW=Facultative wetland and OBL=obligate wetland

majority of which were annual upland Chenopodium opulifolium (Schrad.), species. Chenopodium murale (L.), Taraxacum officinale ((L.) Gaertn.)), Setaria verticillata Leersia hexandra (Sw) species-group ((L.) P. Beauv.)), Cyperus rotundus (L.) however was composed of species that were Palla and Eleusine indica ((L.) Gaertn.). either facultative, facultative wetland or This group was related to vegetation and obligate wetland species. This group was horticultural crops such as cauliflower, also dominated by helophytes and cabbage and, tomato plots within the phanerophytes and were predominantly wetland. perennial weeds. In most cases, they Oxalis corniculata (L) species group were occurred in wet or moist parts of lowland weed species skewed towards moderate floodplains such as Pangani in Tanzania and moisture content. Together, they had group were related to the all season rice growing in members including Commelina Malinda Korogwe. Similar weed benghalensis (L), Centella asiatica (L) composition has been recorded in west Urb)), Galinsoga urticifolia ((Kunth) African rice fields (Kent et al. 2001). Benth.), Vernonia poskeana (Vatke & Malva parviflora (L.) species-group Hildebr.), Kyllinga alata (Nees) and represented mid-montane inland valley weed Stellaria media (L.Vill.). This group was species. These species were mostly exotic dominated by highland inland valley with the majority occurring as either sampling units,, mainly from cropped areas cosmopolitan in distribution or either from a and fallow plots. Other contribution was Mediterranean or a temperate zone. from mid-inland valley. The group was Examples were Chenopodium album (L), associated with phosphorous rich soils and

International Journal of AgriScience Vol. 2(9):810-830, September 2012 823

Fig 8. Raunkiaer’s life forms of weeds in relation to land use gradients in four wetlands, a) highland and lowland floodplains and b) highland and mid-montane inland valleys. FP = floodplain; IV = Inland valley associated with Colocasia esculenta (L.) Plant community composition was biased Schott)) and fallow plots. towards species introductions that were herbaceous therophytes and facultative (FAC) wetlands (Fig. 9). The transition from Weed functional traits relative to a native sedge dominated wetland to one disturbance gradients dominated by grasses and obligate upland

International Journal of AgriScience Vol. 2(9):810-830, September 2012 824

Fig 9. A hierarchical cluster analysis of major land use types relative to Raunkiaer’s life forms, Hydrophytes, Geophytes, helophytes and therophytes annual herbs is indeed a pointer to hydrology, soil and water quality (Ervin et ecological change that has occurred in the al. 2006b). This is also due to the functional composition of a plant characteristically shorter canopy and short community within the small wetlands in life cycle associated with therophytes/and East Africa. Unlike native species, annuals that enable species to grow rapidly introduced species exhibit a greater ability to and complete their life cycles between adjust quickly to a regime shift in disturbed disturbance events (Johnson 1997; Kyle and wetland ecosystems due to such activities as Leishman 2009). In contrast, perennials, draining and partitioning that interfere with typical of un-used or natural wetland

825 International Journal of AgriScience Vol. 2(9):810-830, September 2012 habitats may not complete a life cycle before disturbance and nutrient enrichment, the next wave of disturbance. For instance in whereas native plants adapted to conditions mid-montane inland valleys of Lukozi, the prior to changes are less likely to benefit average maximum fallow period was three (McIntyre et al. 1995; Moles et al. 2008). months; a period that only annuals and Weed invasion threatens wetland biennial weeds can withstand. Soil biodiversity, leading to a decline in both disturbance also tends to favour therophytes species and habitat diversity. A loss of any as this opens up space for colonization functional group may severely affect the through seed establishment. Therophytes are capacity of an ecosystem to reorganize after examples of species that are able to disperse a disturbance (Diaz and Cabido 2001) The effectively and exploit new resources relative decline of other life forms might rapidly (McIntyre et al. 1995) have been compensated for by an increased number of therophytes. Effect of hydric regime on species additions CONCLUSION Plants growing in water are technically Regime shifts in the wetlands of East Africa called hydrophtes. However, not all plants favour species introductions, normally grow in water or water logged soil but some introduced weeds that present a potential also tend to grow in terrestrial habitats threat to biodiversity (Hewitt and Huxel (Tiner 1999). It is noteworthy that reduced 2002). Weed species do increase diversity in flooding induces new species introductions a used wetland however, the fact that they in cultivated wetlands. Facultative upland play similar roles to existing flora creates a weeds completely dominated the drained void in wetland functions that may serve to sites that were studied, as opposed to reduce the ecological integrity of these permanently inundated areas that were wetlands. Even though the study of invasive dominated either by obligate wetland plants is important because they impact on species or by facultative wetland species. A ecological function and structure (Doren et lower hydric regime is conducive for the al. 2009); their indicator capacity is only survival of therophytes and geophyte upland relevant to a specific ecological condition plants that tend to have poor competitive and future restoration success depends on ability in permanently flooded environments the level of invasion and a good (McIntyre et al. 1995). Facultative weed understanding of any adverse impact on an species describe those species adapted to ecosystem. Therefore, there is a need to either seasonal conditions or wet conditions develop simple and cost effective tools to such as those between high moisture identify levels of disturbance in used conditions and low moisture conditions wetlands for all-inclusive and informed (Ervin et al. 2006a). This is also in management decisions. agreement with other investigations demonstrating that those species sensitive to ACKNOWLEDGEMENT permanently inundated conditions such as This work was funded by the Volkswagen Tagetes minuta (L), Oxalis corniculata (L) foundation (Hannover Germany). Thanks to and Bidens pilosa (L) are common in drier Prof. Dr. Salome Misana for facilitating the and nutrient-poor environments (Kercher Small Wetlands of East Africa project in and Zedler 2004). Related to this, introduced Tanzania and for advice. Thanks to species and in most cases invasive plants, Suleiman (TZ), Mr. Mungai and Mr. Musili have been shown to benefit from changes in (KE) for plant identification. Extended environmental conditions such as gratitude to field assistants: Omari

International Journal of AgriScience Vol. 2(9):810-830, September 2012 826

(Korogwe), Hermann (Lushoto), Mzee needs through sustainable wetland use. William (Nyeri), Mzee Koskey, and Gibson Natural Resources Forum 27(2):117- Manyeki (Rumuruti) and Stanley (Tegu- 129 Karatina). Sincere thanks to those Doren RF, Volin JC, Richards JH (2009) anonymous reviewers. Invasive exotic plant indicators for ecosystem restoration: An example REFERENCES from the Everglades restoration Agnew ADQ, and Agnew S (1994) Upland program. Ecological Indicators 9(6, Kenya wild flowers. A flora of the ferns Supplement 1):S 29-S36 and herbaceous flowering plants of Dufrene M, Legendre P (1997) Species upland Kenya. Nairobi: East Africa assemblages and indicator species: the Natural History Society. p 374 need for a flexible asymmetrical ALUKA (2010) ALUKA-Jstor African Plant approach. Ecological Monographs Science database."Retrieved (January 67(3):345-366 2010)" from http://www.plants.jstor.org Ervin G, Herman B, Bried J, Holly D Andreasen C, Stryhn H, Streibig J (1996) (2006a) Evaluating non-native species Decline of the flora in Danish arable and wetland indicator status as fields. Journal of Applied Ecology components of wetlands floristic 3:619-626 assessment. Wetlands 26(4):1114-1129. Chapman L, Balirwa J, Bugenyi F, Ervin G, Smothers M, Holly C, Anderson C, Chapman C, Crisman T (2001) Linville J (2006b) Relative Importance Wetlands of East Africa: biodiversity, of Wetland type Versus Anthropogenic exploitation, and policy perspectives. Activities in Determining Site In: Gopal B, Junk W, and Davis J, Invasibility. Biological Invasions editors. Biodiversity in wetlands: 8(6):1425-1432 assessment, function and conservation. Flinn KM, Lechowicz MJ, Waterway MJ Leiden, the Netherlands: Backhuys (2008) Plant species diversity and Publishers. p 101–131 composition of wetlands within an Daehler CC (1998) The taxonomic upland forest. American Journal of distribution of invasive angiosperm Botany 95(10):1216-1224 plants: Ecological insights and Folke C, Carpenter S, Walker B, Scheffer comparison to agricultural weeds. M, Elmqvist T, Gunderson L, Holling Biological Conservation 84(2):167-180 CS (2004) Regime shifts, resilience and Davis M, Thompson K, Philip Grime J biodiversity in ecosystem management. (2005) Invasibility: the local Annual Review of Ecology, Evolution, mechanism driving community and Systematics 35:557-581 assembly and species diversity. Goettsch B, Hernández H (2006) Beta Ecography 28(5):696-704 diversity and similarity among cactus Diaz S, Cabido M (2001) Vive la difference: assemblages in the Chihuahuan desert. plant functional diversity matters to Journal of Arid Environments ecosystem processes. Trends in 65(4):513-528 Ecology and Evolution 16:646-655 Gordon DR (1998) Effects of invasive, non- Dixon AB, Wood AP (2003) Wetland indigenous plant species on ecosystem cultivation and hydrological processes: Lessons from Florida. management in eastern Africa: Ecological Applications 8(4):975-989 Matching community and hydrological

827 International Journal of AgriScience Vol. 2(9):810-830, September 2012 Haines RW, Lye KA (1983) The sedges and Hulme PE (2006) Beyond control: wider rushes of East Africa : A flora of the implications for the management of families Juncaceae and Cyperaceae in biological invasions. Journal of Applied East Africa, with a particular reference Ecology 43(5):835-847 to Uganda. In: Lye KA (ed) Nairobi: Hyde MA, Wursten B (2011) Flora of East African Natural History Society. p Mozambique data base (retrieved July 404. 2010) from Hall J, Burgess ND, Lovett J, Mbilinyi B, http://www.mozambiqueflora.com Gereau RE (2009) Conservation Johnson DE (1997) Weeds of Rice in West implications of deforestation across an Africa. London, UK: West Africa Rice elevational gradient in the Eastern Arc Development Association (WARDA). p Mountains, Tanzania. Biological 312 Conservation 142(11):2510-2521 Junk W (2002) Long-term environmental Handa C (2011) Vegetation based indicators trends and the future of tropical of used wetlands in East Africa. . Bonn: wetlands. Environmental Conservation Rheinischen Friedrich-Wilhelms- 29(04):414-435 Universität, Bonn. 205 p. Kamiri HWM (2010) Effects of Land Use Harper DM, Mavuti KM, Muchiri SM Dynamics on Attributes of Wetland (1990) Ecology and management of Soils in East Africa. Bonn: Rheinischen lake Naivasha, Kenya, in relation to Friedrich-Wilhelms-Universität. 152 p. climatic change, alien species, Kellman M (1980) Geographic patterning in introductions, and agricultural tropical weed communities and early development. Environmental secondary successions. Biotropica Conservation 17(04):328-336 12(2):34-39 Hewitt C, Huxel G (2002) Invasion success Kent R, Johnson DE, Becker M (2001) The and community resistance in single and influences of cropping system on weed multiple species invasion models: do communities of rice in Côte d'Ivoire, the models support the conclusions? West Africa. Agriculture, Ecosystems Biological invasions 4(3):263-271 & Environment 87(3):299-307 Hill MO (1979) DECORANA - A Kercher S, Zedler J (2004) Flood tolerance FORTRAN program for detrended in wetland angiosperms: a comparison correspondence analysis and reciprocal of invasive and noninvasive species. averaging. Ecology and Systematics Aquatic Botany 80(2):89-102 ((Ithaca, New York) (reissued by Kyle G, Leishman MR (2009) Functional Microcomputerpower, Ithaca).Cornell trait differences between extant exotic, University ). native and extinct native plants in the Hobbs RJ, Mooney HA (1998) Broadening Hunter River, NSW: a potential tool in the Extinction Debate: Population riparian rehabilitation. River Research Deletions and Additions in California and Applications 25(7):892-903 and Western Australia. Conservation Lodge D (1993) Biological invasions: Biology 12(2):271-283 lessons for ecology. Trends in Ecology Howard G, Chege F (2007) Invasions by & Evolution 8(4):133-137 plants in the inland waters and wetlands Lopez R, Fennessy S (2002) Testing the of Africa. Biological invaders in inland floristic quality assessment index as an waters: Profiles, distribution, and indicator of wetland condition. threats:193-208 Ecological Applications 12(2):487-497

International Journal of AgriScience Vol. 2(9):810-830, September 2012 828

Lososová Z, Chytrý M, Cimalová S, Kropá USA: United States Department of Z, Otýpková Z, Pyek P, Tichý L(2004) Agriculture. p 96. Weed vegetation of arable land in Patchett J, Wilhelm G (1999) The Ecology Central Europe: Gradients of diversity and Culture of Water: New Directions and species composition. Journal of for Urban Hydrological Infrastructure Vegetation Science 15(3):415-422 (Revised July, 1999). Citeseer. p 74. Mack JJ (2007) Developing a wetland IBI Perrings C, Mooney HA, Williamson MH with statewide application after (2010) Bioinvasions and globalization : multiple testing iterations. Ecological ecology, economics, management, and Indicators 7(4):864-881 policy. Oxford; New York: Oxford Maclean I, Hassall M, Boar R, Lake I (2006) University Press. Effects of disturbance and habitat loss Pollnac FW, Maxwell BD, Menalled FD on papyrus-dwelling passerines. (2009) Using species-area curves to Biological Conservation 131(3):349- examine weed communities in organic 358 and conventional spring wheat systems. Magee T, Ernst T, Kentula M, Dwire K Weed Science 57(3):241-247 (1999) Floristic comparison of Protabase (2010) Plant resources of tropical freshwater wetlands in an urbanizing Africa database " Retrieved (January environment. Wetlands 19(3):517-534 2010)" http://www.base.prota.org/. McCune B, Grace JB (2002) Analysis of Pyšek P, Jaros ík V, Chytrý M, Kropác Z, Ecological Communities. MjM Tichý L, Wild J (2005) Alien plants in Software Design Gleneden Beach, temperate weed communities: Oregon. p 300 prehistoric and recent invaders occupy McCune B, Mefford M (2006) PC Ord 5.0. different habitats. Ecology 86(3):772- MjM Software, Gleneden Beach. 785 Oregon. p 172 Pyšek P, Richardson DM (2007) Traits McIntyre S, Lavorel S, Tremont RM (1995) associated with invasiveness in alien Plant life-history attributes: Their plants: Where do we stand? In: Nentwig relationship to disturbance response in W (ed.) Biological Invasions: Springer herbaceous vegetation. Journal of Berlin Heidelberg. p 97-125 Ecology 83(1):31-44 Raunkiaer C, Gilbert-Carter H (1937) Plant Moles A, Gruber M, Bonser S (2008) A new life forms. Oxford, UK: Clarendon framework for predicting invasive plant Press. p 104. species. Journal of Ecology 96(1):13-17 Richardson DM, Holmes PM, Esler KJ, Okalebo J, Gathua K, Woomer P (2002) Galatowitsch SM, Stromberg JC, Laboratory methods for soil and plant Kirkman SP, Pyšek P, Hobbs RJ ( analysis: a working manual. 2nd 2007) Riparian vegetation: degradation, edition. The tropical soil biology and alien plant invasions, and restoration fertility program. Regional Office for prospects. Diversity and distributions Science and Technology for Africa. 13(1):126-139 Nairobi, Kenya: UNESCO. Rooney TP, Olden JD, Leach MK, Rogers Olsen SR, Cole CV, Watanabe FS, Dean LA DA (2007) Biotic homogenization and (1954) Estimation of available conservation prioritization. Biological phosphorus in soils by extraction with Conservation 134(3):447-450 sodium bicarbonate. Washington D.C, Sax DF, Stachowicz JJ, Brown JH, Bruno JF, Dawson MN, Gaines SD, Grosberg

829 International Journal of AgriScience Vol. 2(9):810-830, September 2012 RK, Hastings A, Holt RD, Mayfield MM, O'Connor, MI and Rice, William R (2007) Ecological and evolutionary insights from species invasions. Trends in Ecology & Evolution 22(9):465-471 Schuyt KD (2005) Economic consequences of wetland degradation for local populations in Africa. Ecological Economics 53(2):177-190 Stromberg J, Lite S, Marler R, Paradzick C, Shafroth P, Shorrock D, White J, White M ( 2007) Altered stream-flow regimes and invasive plant species: the Tamarix case. Global Ecology and Biogeography 16(3):381-393 Tickner D, Angold P, Gurnell A, Mountford J (2001) Riparian plant invasions: hydrogeomorphological control and ecological impacts. Progress in Physical Geography 25(1):22 Tiner R (1999) Wetland indicators: a guide to wetland indentification, delineation, classification, and mapping. New York: Lewis Publishers. p 392 Turrill W, Beentje H (1952-2007) Flora of tropical East Africa (series). Kew, UK: Royal Botanic Gardens Van der Valk A (1981) Succession in wetlands: a Gleasonian approach. Ecology 62(3):688-696 Violle C, Navas ML, Vile D, Kazakou E, Fortunel C, Hummel I, Garnier E ( 2007) Let the concept of trait be functional! Oikos 116(5):882-892 Wood A, van Halsema G (2008) Scoping agriculture, wetland interactions: Towards a sustainable multiple- response strategy. Rome, Italy: Food and Agriculture Organization of the United Nations. p xviii + 178. Zedler JB, Kercher S (2004) Causes and consequences of invasive plants in wetlands: Opportunities, Opportunists, and Outcomes. Critical Reviews in Plant Sciences 23(5):431 - 452

International Journal of AgriScience Vol. 2(9):810-830, September 2012 830