Rapid morphological change in black rats (Rattus rattus) after an island introduction

Materials and Methods

A total of 61 black rats from Anacapa Island from the Santa Barbara Natural History Museum, the Natural History Museum of Los Angeles County, and the NGO Island Conservation were examined. We received permission from all museums to access the specimens. Two rats were determined to be juvenile by skull suture (DeBlase & Martin, 1974) and excluded from further analysis. The remaining adult specimens and collection years are as follows: 1940 (n = 12), 1975 (5), 1978 (21), 1979 (9), 1986 (2), and 2000 (10). It was not possible to get more modern rat specimens: rats were eradicated on Anacapa in 2001–2002 (Howald et al., 2009). Indeed, the ten year 2000 specimens utilized in this study were collected just prior to the eradication. There were 22 males, 16 females, and 21 unknown.

Eleven cranial measurements were taken following Collins & George (1990) unless otherwise indicated. Measurements included: alimentary toothrow (AL), breadth of braincase (BB), breadth of rostrum (BR), depth of braincase (DBC), greatest length of skull (GL), interorbital breadth (IB), length of braincase (LBC), length of incisive foramen (LIF), length of palate plus incisor (LPN, measured as the greatest distance from the end of the nasals to the mesopterygoid fossa), length from supraorbitals to nasals (ONL, measured as the least distance from the supraorbital notch to the tip of the nasals), and zygomatic breadth (ZB).

Cranial measurements of a total of 59 specimens were taken by DB, RJ, and KL with digital calipers to the nearest 0.5 mm. Each worker measured each trait three times and utilized the mean. A threshold variance > 0.1 of the three means required re-measuring of that trait. The four standard external measurements in museum specimens were originally made by numerous different museum preparers and recorded from museum tags: total length (TOT), tail length (TAIL), hind foot length (HF), and ear length (EAR). Because of either lack of external measurement by museum preparers or damage to the skulls, some measurements were not available for some specimens. In particular, we did not have external measurements for 23 specimens. In 13 cases this was due to skulls resident in museum collections not having external measurements attached. The 10 specimens from Island Conservation were so highly decayed that any external measurements would have been guesswork. In all cases all measurements available to us were used.

We used visual examination of normal probability plots (Afifi, Clark & May, 2004) and the Shapiro–Wilk W statistic to test for normality of distribution. The Kolmogorov–Smirnov D statistic test was not used because these tests have poor power properties and tend to reject the null hypothesis with large sample size and accept it with small sample size (Afifi & Clark, 1997).

To evaluate sexual dimorphism, two-sample t-tests were performed after dropping 21 specimens of unknown sex (leaving 38 specimens of known sex). Results of t-tests were considered significant at the 95% confidence level. The specimens of unknown sex were included in subsequent analyses.

We used two methods to evaluate changes over time: (A) categorical analysis and (B) linear regression Because the longest gap between collection years was 35 years, and the next longest gap was only 14 years, we chose to perform our categorical analysis (MANOVA) on two time periods = 1940 & 1975–2000. Also, these categories fit well with when Anacapa deer mice were collected (1940 & 1978), allowing for more direct comparison. However, damage to specimens and lack of measurements did not permit inclusion of the full multivariate data on many of the specimens. Accordingly, we also performed independent-samples t-tests, testing the significance of the difference between the sample means of the pre- and post-1950 time periods of each measurement of all specimens. We used the Levene statistic (Brown & Forsythe, 1974) to test the assumption of equal variance, and applied the appropriate t-test. Although there have been recent and substantial objections to the use of sequential Bonferroni corrections (especially by ecologists; Moran, 2003), to be conservative we then applied a Holm–Bonferroni sequential correction to account for the multiple tests being conducted (Holm, 1979). To compare means of non-normally distributed traits, we used Kruskal–Wallis H tests.

Because we wished to evaluate rates of change as well as total amounts of change, we calculated the rate of annual change in each significant trait. This was done by dividing the difference in means by the difference in the means of collection years, both between periods. Darwins were calculated with the equation: ${\displaystyle d=|\left(ln{x}_2-ln{x}_1\right)/\left(t_2-t_1\right)|,}$ where lnx₁ and lnx₂ are sample means of ln measurements at times t₁ and t₂, respectively (measured in millions of years).

However, evolution calculated in standard deviations per generation (haldanes) has several advantages. Rates in haldanes are independent of dimension in a way that rates in darwins are not, since traits are passed onto progeny only between generations, and generation times vary between organisms by many orders of magnitude. This makes haldanes more readily comparable in terms of quantitative evolutionary genetics (Gingerich, 1993). Haldanes were calculated with the equation: ${\displaystyle h=|\left[\left(ln{x}_2/\mathrm{sln}x\right)-\left(ln{x}_1/\mathrm{sln}x\right)\right]/\left(t_2-t_1\right)|,}$ where ln x₁ and ln x₂ are sample means of ln measurements at times t₁ and t₂ respectively (measured in generations), and sln x is the pooled standard deviation of ln x₁ and ln x₂ (Haldane, 1949). Published demographic data was used to estimate generations per year (Erickson & Halvorsen, 1990).

However, the two time periods compared a very small early (1940) sample (only 12 of the 59 specimens). To lessen this effect, we performed linear regressions comparing each measurement of each trait, with the year in which the specimen was collected.

Results

Data were first examined for normality of distribution through inspection of normal probability plots (Afifi, Clark & May, 2004) and Lilliefors test (SPSS, 1998). We found all traits except AL to be normally distributed. There was no significant sexual dimorphism for any of the 15 measurements.

Results of Levene’s tests, independent samples t-tests, and Kruskal–Wallis tests are given in Table 1. In rats, all cranial measurements (but no external measurements) were found to have changed over time. We feel our not finding significant change in external measurements was very likely due to insufficient sample size; missing from 23/59 (39%) of the specimens. MANOVAs corroborated individual results from the t-tests. Rats collected in 1940 were extremely significantly different from rats collected in 1975–2000 (Wilk’s λ < 0.0005, F_d.f.15 = 2974.386, P < 0.0005).

Table 2 gives amounts and rates of evolutionary change. All rat cranial traits changed 3.06–10.43% (724–2567 d, 0.06–0.43 h), and all became larger. All external traits also grew larger, but probably because of much smaller sample size, not significantly so.

Table 3 and Fig. 1 show the results of our linear regressions. All cranial traits were found to be significant, while none of the external traits were found to be significant.

Discussion

All rat cranial traits changed, some dramatically, and all became larger. When considered in the more accurate haldanes, some of these changes are among the fastest on record (Hendry, Farrugia & Kinnison, 2007), the maximum of 0.43 h (trait IB) exceeded only by systems such as Trinidadian guppy spot number, spot area, spot length, offspring size, & size: 0.43–0.74 h (Endler, 1980; Reznick et al., 1997); Galapagos finches weight, bill depth, bill width, bill length, beak size: 0.43–0.71 h: (Grant & Grant, 1995; Grant & Grant, 2002); and freshwater copepod egg type switch date: 0.49 h (Hairston Jr& Walton, 1986). These rates of change in the rats are far above the highs found in other island rodents, including invasive black rats on the Galapagos Islands and endemic deer mice on the California Channel Islands [0.03 h max (Pergams & Ashley, 2001)].

It is interesting that our linear regressions showed all cranial traits to be significant, and that all cranial traits increased monotonically across the three time periods. Four possible explanations for temporal variation of phenotypic characters are: (1) nongenetic, environmental effects (plasticity); (2) gene flow from morphologically different source populations; (3) stochastic evolutionary change through genetic drift; or (4) response to natural selection.

We feel that non-genetic environmental factors such as nutrition or maternal effects are an unlikely explanation for all of the morphological changes observed. First, changes appear so great that some genetic component must be involved. Second, all of the rat’s cranial measures increased in size, while at largely the same time most of the mice’s traits decreased in size. If environmental plasticity were the cause of the changes we would expect both species’ traits to change in the same direction.

Several factors argue against gene flow in the case presented here. First, the rats inhabit oceanic islands separated by at least several kilometers from other islands or from the nearest mainland point. Second, there is no record of rats on nearby islands.

A third causative factor that merits consideration is genetic drift. Anacapa is quite small (2.9 sq km, all Anacapa islets together), but mice are often extremely abundant. A three to four year population size cycle is thought to exist on Anacapa, but the minimum size is still probably in the thousands (Pergams, Lacy & Ashley, 2000). A final argument against genetic drift is the observation that in both rats and mice many morphological characters are changing, and in most cases changes are in the same direction. Even if certain characters exhibited higher levels of evolutionary plasticity, genetic drift would not be expected to direct changes in many such characters in a uniform direction.

Over approximately the same time frame that some rat traits were getting bigger, endemic deer mice were mostly getting smaller, except that deer mouse ear length and nostral width got bigger (Pergams & Ashley, 1999; Pergams & Ashley, 2000; Pergams & Ashley, 2001). The changes in these mice do not seem to be due to climate change or changes in human population density (Pergams & Lawler, 2009), factors which were evaluated in that paper. The changes toward smaller size in the mice are also not consistent with release of selective pressures upon introduction.

We speculate that changes towards mostly smaller size in mice may have been driven by competition with and predation by black rats. Collins, Storrer & Rindlaub (1979) study stomach content of both species, and find that both eat at least seven of the same plants: sand lettuce (Dudleya caespitosa), wild cucumber (Echinocystis lobate), coastal prickly pear (Opuntia littoralis), sea fig (Carpobrotus chilensis), iceplant (Carpobrotus edulis), slender-leaved iceplant (Mesembryanthemum nodiflorum), and holly-leafed cherry (Prunus ilicifolia). Although rats do not eat giant coreopsis (Coreopsis gigantean) as mice do, rats and mice both employ this large-leaved plant as shelter and cover and so may compete for it. Direct predation of rats upon mice is also a consideration: rats on Anacapa prey on mouse pups, as well as on mice caught in traps (Collins, Storrer & Rindlaub, 1979).

If rats were a factor in mouse evolution on Anacapa, it would be interesting to evaluate mouse morphology after the rat eradication in 2001–2002. Such future research might sample Anacapa mice periodically (perhaps every 5–15 years) and begin ongoing analysis after perhaps 20–30 years. If mice were generally getting bigger again, it would further support the hypothesis of competition with and/or predation by rats as a proximate cause of mouse evolution on Anacapa Island.

Supplemental Information